Mashape's Voice

Lemma what? A guide to Text Processing and Machine Learning API terms

Fri, 17 May 2013 11:28:00 -0400

After we posted the a list of NLP, Sentiment Analysis, and Machine Learning APIs a while ago, we noticed that some API descriptions require a little bit of digging into, to fully appreciate what these APIs can do. Here’s an example:

Text analysis API including wordnet synsets,relation extraction,named entity recognition and classification,lemmatization,part of speech tagging,tokenization, and semantic role labeling.

If you’re not familiar with these words, you could totally miss the features that this API is capable of.

To help with that, we have listed below an explanation to some of these words in the NLP/Machine Learning context; as well as APIs (represented as numbered links) whose descriptions mention these terms. Hopefully a basic understanding of these terms would help you appreciate what these APIs are capable of.

Stemming and Lemmatization (1, 2, 3)

Stemming is the process of removing and replacing word suffixes to arrive at a common root for of the word. Lemmas differ from stems in that a lemma is a canonical form of the word, while a stem may not be a real word. (Reference)

For example, from “produced”, the lemma is “produce”, but the stem is “produc-“. This is because there are words such as production. (Reference)

Text Analytics (1, 2, 3, 4, 5, 6)

Describes a set of linguistic, statistical, and machine learning techniques that model and structure the information content of textual sources for business intelligence, exploratory data analysis, research, or investigation. (Reference)

Rough equivalent to the term “text data mining”, it usually involves the process of structuring input text (usually parsing), deriving patterns within the structured data, and finally evaluation and interpretation of the output. Typical text mining tasks include text categorization, text clustering, concept/entity extraction, production of granular taxonomies, sentiment analysis, document summarization, and entity relation modeling (i.e. learning relations between named entities). (Reference)

Tokenization (1)

Is the process of breaking a stream of text up into words, phrases, symbols, or other meaningful elements called tokens. The list of tokens becomes input for further processing such as parsing or text mining. (Reference)

Named-entity recognition/extraction (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)

Seeks to locate and classify atomic elements in text into predefined categories such as the names of persons, organizations, locations, expressions of times, quantities, monetary values, percentages, etc. (Reference)

Example:

Jim bought 300 shares of Acme Corp. in 2006.

Producing an annotated block of text, such as this one:

<ENAMEX TYPE=”PERSON”>Jim</ENAMEX>bought<NUMEX TYPE=”QUANTITY”>300</NUMEX>shares of<ENAMEX TYPE=”ORGANIZATION”>Acme Corp.</ENAMEX> in <TIMEX TYPE=”DATE”>2006</TIMEX>.

Sentiment Analysis (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)

Aims to determine the attitude of a speaker or writer with respect to some topic or the overall contextual polarity of a document. A basic task in sentiment analysis is classifying the polarity of a given text at the document, sentence, or feature/aspect level - whether the expressed opinion in a document, a sentence, or an entity is positive, negative, or neutral. (Reference)

Summarization (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)

Process of reducing a text document with a computer program in order to create a summary that retains most important points of the original document. (Reference)

Chunking (1, 2, 3)

Also called “light parsing”, is an analysis of a sentence which identifies the constituents (noun, verbs), but does not specify their internal structure, nor their role in the main sentence. It is similar to the concept of lexical analysis for computer languages. (Reference)

(Word-sense) Disambiguation (1, 2, 3, 4)

The process of identifying which sense of a word (meaning) is used in a sentence, when the word has multiple meanings.

Consider the word “bass” which could either mean “a type of fish” or “tones of low frequency”, depending on the sentence it was used on. (Reference)

Part-of-speech tagging (1, 2, 3, 4)

(POS) Also called grammatical tagging or word-category disambiguation, is the process of marking up a word in a text (corpus) as corresponding to a particular part of speech, based on both its definition, as well as its context - i.e. relationship with adjacent and related words in a phrase, sentence, or paragraph. (Reference)

To check out the difference between chunking and POS, check here.

Semantic role labeling (1)

Also sometimes called shallow semantic parsing, is a task in natural language processing consisting of the detection of the semantic arguments associated with the predicate or verb of a sentence and their classification into their specific roles. For example, given a sentence like “Mary sold the book to John”, the task would be to recognize the verb “to sell” as representing the predicate, “Mary” as representing the seller (agent), “the book” as representing the good (theme), and “John as representing the recipient. (Reference)

Collaborative filtering

In the newer, narrower sense, collaborative filtering is a method of making automatic predictions (filtering) about the interests of a user by collecting preferences or taste information from many users (collaborating). The underlying assumption of the collaborative filtering approach is that if person A has the same opinion as a person B on an issue, A is more likely to have B’s opinion on a different issue x than to have the opinion on x of a person chosen randomly. For example, a collaborative filtering recommendation system for television tastes could make predictions about which television show a user should like given a partial list of that user’s tastes (likes or dislikes). Note that these predictions are specific to the user, but use information gleaned from many users. This differs from the simpler approach of giving an average score for each item of interest, for example based on its number of votes.

Cluster analysis or clustering

Is the task of grouping a set of objects in such a way that objects in the same group (called cluster) are more similar (in some sense or another) to each other than those in other groups (clusters). It is a main task of exploratory data mining, and a common technique for statistical data analysis used in many fields, including machine learning, pattern recognition, image analysis, information retrieval, and bioinformatics.

Popular notions of clusters include groups with small distances among the cluster members, dense areas of the data space, intervals, or particular statistical distributions. (Reference) Also see “Classification” below.

Classification (1, 2, 3, 4, 5, 6, 7, 8)

Is the problem of identifying to which of a set of categories a new observation belons, on the basis of a training set of data containing observations who category membership is unknown. Classification is considered an instance of supervised learning i.e. learning where a training set of correctly-identified observations is available. The corresponding unsupervised procedure is known as clustering (cluster analysis), and involves grouping data into categories based on some measure of inherent similarity. (Reference)

Supervised versus Unsupervised Learning

Machine learning algorithms are described as either ‘supervised’ or ‘unsupervised’. The distinction is drawn from how the learner classifies data. In supervised algorithms, the classes are predetermined. These classes can be conceived of as a finite set, previously arrived at by a human. In practice, a certain segment of data will be labelled with these classifications. The machine learner’s task is to search for patterns and construct mathematical models. These models are then evaluated on the basis of their predictive capability in relation to measures of variance itself. Decision tree induction and naive Bayes are some examples of supervised learning techniques.

Unsupervised learners are not provided with classifications. Unsupervised algorithms seek out similarity between pieces of data in order to determine whether they can be characterized as forming a group. These groups are termed clusters, and there are a whole family of clustering machine learnign techniques. (Reference)

Recommender system (1, 2, 3)

A subclass of information filtering system that seek to predict the ‘rating’ or ‘preference’ that user would give to an item (such as music, books, or movies) or social element (people or groups) they had not yet considered, using a model built from the characteristics of an item (content-based approaches) or the user’s social environment (collaborative filtering approaches. (Reference)

Recommender systems have become extremely common in recent years. A few examples of such systems:

When viewing a product on Amazon.com, the store will recommend additional items based on a matrix of what other shoppers bought along with the currently selected item.^[3]
Pandora Radio takes an initial input of a song or musician and plays music with similar characteristics (based on a series of keywords attributed to the inputted artist or piece of music). The stations created by Pandora can be refined through user feedback (emphasizing or deemphasizing certain characteristics).
Netflix offers predictions of movies that a user might like to watch based on the user’s previous ratings and watching habits (as compared to the behavior of other users), also taking into account the characteristics (such as the genre) of the film.

Neural Networks

A computing system made up of a number of simple, highly interconnected processing elements, which process information by their dynamic state response to external inputs. (Reference)

The power of neural networks is that it can make reasonable guesses about results for queries it has never seen before, based on similarity to other queries. For example, search engines can use neural network algorithms to provide best guess answers to queries that have not been typed in before.

——

If there are other terms you’d like to add here, or APIs that should be associated with a term, let chris@mashape.com know.

New Feature: Secret Parameter and Headers

Thu, 16 May 2013 14:33:00 -0400

It’s been a while since we announced a new feature from Mashape, so we’re very excited to announce this one: Secret Parameters and Headers for API providers.

What is a Secret Parameter/Header?

A Secret Parameter or Header are fixed key-value that is added to the request coming from the Mashape proxy to the API provider. Only API providers can set these values in their API.

What is it useful for?

One particular useful use-case for them is where developers have to sign up for separate accounts from the API provider and Mashape. Having these values allow API providers to set a dedicated pre-registered account that can represent all users coming from Mashape to their API; or as an alternative identifying method. These users can be further identified using the X-Mashape-User header that the API receives from Mashape.

The effect is a more streamlined process of API registration, testing, and consumption. This saves developers the extra step of signing up for several accounts.

You can set your Secret Parameters and Headers in your API Admin page, under the Proxy tab. See video below:

We are encouraging all API providers to use this feature. Let us know what you think! - support@mashape.com

Upcoming Hackathons: iCreate Mobility Challenge + APIDays Mediterranea

Tue, 14 May 2013 18:41:00 -0400

Mashape is taking part in two upcoming hackathons. We would like to thank our API partners who are offering free access or mini-prizes for the iCreate Mobility Challenge 2013 (Singapore May 17th) and APIDays Mediterranea (Madrid, May 30). See bottom for the list of APIs and their offers.

For the hackathon organizers: please consolidate your participants’ Mashape IDs and send them a message in Mashape BEFORE your hackathon. This will give them ample time to enable access to their private APIs to your participants.

If you are a Mashape API partner and interested to offer free access to your paid API or provide a mini-prize, please email chris@mashape.com

Good luck to all participants!

Participating API partners and their offers:

Text-Processing - Sentiment analysis, stemming and lemmatization, part-of-speech tagging and chunking, phrase extraction and named entity recognition. (Offer: Free access to API during the hackathon)

Diffbot - Diffbot uses computer vision and natural language processing to extract data from web pages automatically. For example, the Article API will return the title, author, date, images and full-text from any blog post or news article web page. Diffbot also offers API-on-the-fly functionality that allows you to extract whatever data you choose from practically any site. Use the web as your database! (Offer: Free access to API during the hackathon when “APIDays” or “iCreate” is stated as reason in registration process).

PeerReach - The PeerReach API allows you to give context to content produced by users. Currently we only support Twitter users but will accept other networks in the near future. (Offer: TBA)

SkyBiometry Face Detection and Recognition - An easy to use Face Detection and Recognition API. (Offer: Get unlock code from your hackathon organizer. Once you have received the unlock code, follow the steps)

Plans:

Valid 14 days after start of the respective event regardless of activation time

5000 API calls, 1000 hourly limit, 5000 daily limit

Instructions to unlock special plans for the hackathon:

Sign or create account (additionally to Mashape account) at https://skybiometry.com/Account/Login

After login, go to “Account” page and under “Subscription” section select “upgrade or change”

After “Select subscription” page appears, enter unlock code to “Special offer unlock code:” field and press “Unlock” button

After additional subscription plan appears, press “Upgrade” button under it

Press “Confirm” button in popped up dialog box

Portrait3D - Portrait3D API is based on Avatarion’s Tethys 3D™ solution, and provides software developers with a technology to create animated facial models based on photos. (Offer: Free access to private API during hackathon)

Algorithms.io - The Algorithms.io API provides a catalog of machine learning algorithms as a service. Includes recommendation algorithms (collaborative filtering), clustering, and classification. Check back with us often as we are constantly adding new algorithms. If you have a machine learning algorithm you’d like to offer for use via our API please contact us to become part of our partner program. You can find detailed documentation on our docs page at: http://documentation.algorithms.io (Offer: Free access to API during hackathon, more details TBA)

Speech2Topics - Yactraq Speech2Topics is a cloud service that converts audiovisual content into topic metadata via speech recognition & natural language processing. Customers use Yactraq metadata to target ads, build UX features like content search/discovery and mine Youtube videos for brand sentiment. In the past such services have been expensive and only used by large video publishers. The unique thing about Yactraq is we deliver our service at a price any product developer can afford. (Offer: Free access to private API during hackathon)

DuckDuckGo Zero-click Info - DuckDuckGo Zero-click Info includes topic summaries, categories, disambiguation, official sites, !bang redirects, definitions and more. You can use this API for many things, e.g. define people, places, things, words and concepts; provides direct links to other services (via !bang syntax); list related topics; and gives official sites when available. Best of all, searches on DuckDuckGo are entirely private because DuckDuckGo doesn’t collect or share any personally identifiable information. (Offer: TBA)

Hoppit Restaurant Guide - Dubbed as “Yelp meets Pandora,” the Hoppit API answers the ongoing question “Where should we eat tonight?” by giving you access to the restaurant data used by our award-winning applications, as seen on FOX, CBS, and Mashable. This includes the nation’s largest repository of high-resolution restaurant photography and “Vibe” data, available for 20 U.S. cities. (Offer: Free access to private API during hackathon)

SpringSense Meaning Recognition - The fastest and most accurate Meaning Recognition (Word Sense Disambiguation) API in the world. Recognises any nouns in a body of text and allows you to provide a rich user-interface with meaning definitions. (Offer: Free access to private API during hackathon)

List of fun and interesting APIs to try out!

Thu, 09 May 2013 20:05:52 -0400

Here are some APIs that crack us up here at Mashape, as well as those that we like to show off to friends :)

face - Fuse two faces into one. That’s very FUN!
Yoda Speak - Turn your sentences and webpage into Yoda-speak!
ERMAHGERD Translator - This translates text to ERMAHGERD.
Gender Guesser - Guess whether a first name is of male or female gender in the USA.
l33t sp34k - Want to sound like a hacker? Translate text to and from l33t sp34k!111eleven!!!
Text to Voice - reads out your text, and in different languages too
Text to Speech - A really simple api that lets you convert text to speech. It’s 100% free for unlimited usage.
Text To Speech - Turns text into an mp3 audio file with a nice female voice similar to Siri.
Text-to-speech - This API instantly generates high quality speech from text. Simply send a request with the text you want read, and we’ll instantly return a wav, mp3, or ogg with it been spoken!
Text-to-Speech - The Voice RSS Text-to-Speech (TTS) API allows conversion of textual content to speech easier than ever. Just connect to our Text-to-Speech (TTS) API with a few lines of code and get verbal representation of a textual content.
Numbers - An API for interesting facts about numbers. Provides trivia, math, date, and year facts about numbers. For example, “5 is the number of platonic solids”, “42 is the number of little squares forming the left side trail of Microsoft’s Windows 98 logo”, “February 27th is the day in 1964 that the government of Italy asks for help to keep the Leaning Tower of Pisa from toppling over”
Jeannie - Jeannie (Voice Actions) is a virtual assistant with over two Million downloads, now also available via API. The objective of this service is to provide you and your robot with the smartest answer to any natural language question, just like Siri.
Trivia - The Trivia API gives you access to 1000s of Trivia questions. New ones are being added all the time! The database is also being expanded with new categories and feedback mechanisms coming soon!!
EgyptoName - EgyptoName translate your name and create a beautiful image using the Egyptian Hieroglyphics.
Texas Holdem - With this API you can easily create a full Texas Holdem game, just create the graphic and associate it with the results.
Word Cloud Maker - Generate Word Clouds from blocks of text. Multi color and different sizes illustrate the frequency, and the “vibe” of the bigger text. Try it with blog posts, article text, speeches, tweet histories, emails, or any other written word.
Bingo - With this API you can easily develop a bingo game, just design the graphic. - Card’s makers - Extraction number - Winner cards - Points

Mashape is a Finalist for the 2013 Red Herring Top 100 North America Award

Wed, 08 May 2013 13:01:00 -0400

San Francisco - May 8, 2013 – Mashape announced today it had been selected as a finalist for Red Herring’s Top 100 North America award, a prestigious list honoring the year’s most promising private technology ventures from the North American business region.

Red Herring has been selecting the most exciting and promising start-ups and “scale ups” since 1995. Finalists are still evaluated individually from a large pool of hundreds of candidates based across North America. Twenty major criteria underlie the scoring and process. They include, among others: the candidate company’s addressable market size, its IP and patents, its financing, the proof of concept, trailing revenues and management’s expertise. Each company goes through an individual interview after filling out a thorough submission, complemented by a due diligence. The list of finalists often includes the best performing and prominent companies of that year.

This unique assessment of potential is complemented by a review of the company’s actual track record and standing, which allows Red Herring to see past the “buzz” and make the list a valuable instrument for discovering and advocating the greatest business opportunities in the industry.

2013 will be remembered as a special vintage. “The finalists list confirms the excellent choices made by entrepreneurs and VCs and the start-ups’ solid roots in corporate America, embracing their innovations. By all metrics, it emphasizes the United States’ entrepreneurial excellence,” said Alex Vieux, publisher and CEO of Red Herring.

Finalist selections for the 2013 edition of the Red Herring 100 North America award are based upon technological innovation, management strength, market size, investor record, customer acquisition and financial health. During the several months leading up to the announcement, hundreds of companies in the fields of security, Web 2.0, software, hardware, life sciences, cloud, mobile and others completed their submissions to qualify for the award.

Finalists are asked to present their winning strategies at the Red Herring North America Forum in Monterey, Calif., May 21 to 23, 2013. The Top 100 winners will be announced at a special awards ceremony the evening of May 23 at the event.

Please follow the Red Herring conference at https://twitter.com/digitalherring hashtag #RedHerring100

Web-enable your Research/Project with an API

Tue, 30 Apr 2013 18:21:00 -0400

This post is intended to help data scientists and engineers who in some capacity have implemented routines/algorithms/data that does a specialized function (e.g. machine learning) using a dynamically typed language, such as Python. The goal is to web-enable these routines/algorithms using an application programming interface (API).

Exposing these functions/data as an API allows for:

Easier, faster, and consistent sharing of functionality/data that could further progress the research. A good example of this is the Materials Project from the Lawrence Berkeley National Laboratory. They deemed it necessary for the scientific community to have access to their data, hence exposing it as an API.
Generating revenue, either as a business entity (e.g. university spinout) or simply to cover the costs of running the API infrastructure. Two cases that come to mind are a) Chatterbox and b) Text Processing, respectively. You can also check out a list of 40+ Machine Learning APIs (some require a paid subscription)

Another general reason for implementing an API layer on your functionality/data is that the web technology and the available libraries and tools have progressed to a point where complexity is kept at a minimum e.g. enabling one to pick and choose their preferred technology stack with ease.

For this post, we will use Python as the programming language because aside from its deep history with the scientific community, it has several scientific and numeric packages available for it (e.g. NumPy, SciPy, etc). If you’re already using some of these, then this post applies to you (even if we’re not going to use these libraries). If Python is not your cup of tea, rest assured that the concepts described here apply all the same. The approach/steps below assume that you could be coming from different languages/platforms.

Let’s get started!

To add an API layer to your project, we would need three main ingredients: a web hosting, your project, and a framework/library to implement a REST API. Let’s represent them as diagrams so you can visualize how the will fit together.

The final implementation will look like the following:

What is REST?

REST stands for Representational State Transfer. It is a web “architecture” that lets us access/modify web resources through HTTP (yes similar to the browser “http://” prefix). For example, under the REST architecture, your endpoint ideally should be accessible using a browser as “http://yourdomain.org/yourproject/endpoint?parameter=value”. This is an alternative to web services (SOAP, XML, etc). For most programming languages, there is a REST framework available.

4 Steps to gathering and putting this all together:

Searching for web hosting
Searching for a REST framework library
Choosing your endpoints
Putting it all together (if you’re pressed for time, you can go here directly and skip steps 1-3)

1. Searching for web hosting

Web hosting is a service that allows us to host our application to be served over the web. The server where our application resides will “serve” up requests from external people/software that want to access our API. Depending on the language you used for your application, there are usually free or cheap options for web hosting. In most cases you’d probably want to try a Freemium model where usage have limits (e.g. 500MB storage, 1GHz processor, etc). There are also “cloud” options where you can “spin” up virtual machines on demand (useful for cases where you’re not expecting demand for your API to spike up as a result of being featured in a popular syndication site like kdnuggets.com). These cloud options are usually better because they have additional bells and whistles such as an integrated development environment

Since we can’t account for all permutations of programming languages, hosting, and cloud options out there, you could cut through the Google search and go straight to Quora.

Quora is a question and answer website moderated by a community of users. With Quora, people can “vote up” answers they like, eventually putting the best answer at the top. But more important, these answers have additional commentary that provides context. Here’s a search for “python hosting” - http://www.quora.com/search?q=python+hosting (Another site to check out is Stackoverflow, which caters to more technical and specific programming questions)

The last result at the bottom seems relevant to us - http://www.quora.com/Python-programming-language-1/Where-do-you-host-your-Python-based-web-apps

There’s no silver bullet to choosing the right hosting, that is unless you test them all out individually. In some cases, you don’t have to deploy a fully working application to decide whether it’s the right platform/hosting or not. If it’s starting to “get in the way” (e.g. getting too many errors, few support systems, etc.) while you’re trying to do something, then it’s safe to try another one. It’s that simple.

This is of course biased, given that everyone will have different levels of technical background, experience, etc. The key thing here is to choose the one that helps you.

In my case, I eventually chose PythonAnywhere simply because they have this hovering help box that guides you throughout the process of deploying a “Hello World” application:

What is Hello World?

Hello World is a term that refers to the simplest code one could ever write on a specific language to get it to print out “Hello World”. It is a good way to introduce novice programmers trying out a new language or platform.

2. Searching for a REST framework library

Now that we know about Quora, this should be quick - http://www.quora.com/What-is-a-good-Python-framework-for-building-a-RESTful-API

The majority seems to like “flask”. It just so happens that PythonAnywhere supports it:

We’re in luck! (Actually I did the two Quora searches above simultaneously before I decided on a combination of PythonAnywhere and Flask. So I’m cheating. But you get the point :P).

3. Choosing your application (endpoints)

If you already have an application that you want to put an API layer on, then you don’t have to search for a project. In my case though, I had to find an example that I can use for this post.

You do however have to figure out the application functionality or data that you want to expose as an API endpoint. This will dictate the REST structure of your endpoint. Let’s use an example Python application I picked to demonstrate this.

Just a few days ago someone in Hacker News posted an algorithm for summarizing paragraphs of text using Python. I figured the programming world can benefit from it if the algorithm is exposed as an API. But what information does my API need to summarizing text? Looking at the source code snippet for the application:

The main SummaryTool object accepts “content” and consequently a “title”. As a REST endpoint, this could look like “http://yourdomain.org/yourproject/summarize?title=<title here>&content=<content here>”

4. Putting it all together

I would assume that a majority of you at this point have only been reading and haven’t actually signed up/tried the service/library I mentioned above. That’s ok. Now’s the time to do that.

Let’s start with PythonAnywhere and Flask:

a. Go to https://www.pythonanywhere.com and sign up for a “Noob” account. Like I mentioned above, you can try out these services first and decide later if you want to commit.

After signing up choose “I want to create a web application”

b. It’s pretty much smooth-sailing from here. Just follow the instructions at the green box at the top that will show you how you can create your Hello World application. We will build on top of this Hello World app later to import the Flask library and add the Summarization code.

There are two interesting points here that you need to be mindful of during this process:

Your domain will look similar to this:

It’s ok *not* to upgrade at this point.

Next is the framework. Click on “Flask”, then click Next to choose the default file name for your project.

It will take a few minutes to spin up your new application, but that should be all you have to do to have a working Hello World Python running on PythonAnywhere with flask.

c. You can go to the browser to access it:

Now let’s integrate our Summarization code with this new Hello World app. I’ve pasted the code here for your convenience.

But here are the relevant parts that were added:

The first yellow box at the top highlights the libraries required by the Summarization algorithm (copied over at the bottom in the picture), plus the urllib2. The 2nd yellow box highlights imports that are required by our additional code highlighted in the 3rd yellow box.

The 3rd yellow box highlights the main addition to the original Hello World code, which describes the new endpoint route “/summarize” accepting two parameters, “title” and “content”. We then pass them to the SummaryTool object, and the results are sent back as JSON to the calling application.

What is JSON?

JSON stands for Javascript Object Notation. It is simply a format for exchanging data across applications. The other alternative is XML. Check out this Stackoverflow question on why choose JSON.

Once you’ve edited your Hello World application like this, make sure to save “Reload” your application to apply the changes. You can reload your app from the Dashboard -> Web tab:

Our API is now ready!

Try it out by clicking this . You should get a summarized output based on the title and content provided to the API.

—

To recap: We have API-enabled our Python app using a web hosting/cloud service, and a REST framework library. This will make it accessible to the rest of the software community and hopefully further progress of its research and foster innovative implementations/mashups.

There are other considerations not covered in this post such as security (OAuth, HTTP Basic, Query key parameters, etc), scalability, rate-limiting/throttling, and other API “to-dos”. I would suggest that you check this exhaustive list of 43 Things to think about when Designing, Testing, and Releasing your API. There’s quite a number of things to check there. Your API journey has just begun!

Bonus step:

I’ve saved the best step for last - promoting your API to developers through Mashape.

Mashape is a a marketplace where developers can discover, distribute, and monetize private and public APIs. We make it easy for you to make your API known to the developer world. It’s free to join and add your API to Mashape. You can check out these tutorials on how you can add your API to Mashape - https://www.mashape.com/docs/gettingstarted/tutorials

In fact, I put the API we created here up on Mashape. You can test it here - https://www.mashape.com/ismaelc/summarizer-tool

Once your API is up on Mashape, it is just as easy to start adding pricing plans to your API, like this Social Media Sentiment Analysis API from Chatterbox.

Hope this post inspired/help you to turn your research/projects/apps into APIs. If you have questions or ideas for future posts feel free to email me at chris@mashape.com

15 Natural Language Processing APIs

Fri, 26 Apr 2013 15:35:00 -0400

Natural Language Processing, or NLP, is a field of computer science, artificial intelligence, and linguistics concerned with the interactions between computers and human (natural) languages.

Here are useful APIs that help bridge the human-computer interaction:

Text Processing - The WebKnox text processing API lets you process (natural) language texts. You can detect the text’s language, the quality of the writing, find entity mentions, tag part-of-speech, extract dates, extract locations, or determine the sentiment of the text.
Question-Answering - The WebKnox question-answering API allows you to find answers to natural language questions. These questions can be factual such as “What is the capital of Australia” or more complex.
Jeannie - Jeannie (Voice Actions) is a virtual assistant with over two Million downloads, now also available via API. The objective of this service is to provide you and your robot with the smartest answer to any natural language question, just like Siri.
Diffbot - Diffbot extracts data from web pages automatically and returns structured JSON. For example, our Article API returns an article’s title, author, date and full-text. Use the web as your database! We use computer vision, machine learning and natural language processing to add structure to just about any web page.
nlpTools - Text processing framework to analyse Natural Language. It is especially focused on text classification and sentiment analysis of online news media (general-purpose, multiple topics).
Speech2Topics - Yactraq Speech2Topics is a cloud service that converts audiovisual content into topic metadata via speech recognition & natural language processing. Customers use Yactraq metadata to target ads, build UX features like content search/discovery and mine Youtube videos for brand sentiment.
Stremor Automated Summary and Abstract Generator - Language Heuristics goes a step beyond Natural Language Processing to extract intent from text. Summaries are created through extraction, but maintain readability by keeping sentence dependencies intact.
Repustate Sentiment and Social Media Analytics - Repustate’s sentiment analysis and social media analytics API allows you to extract key words and phrases and determine social media sentiment in one of many languages. These languages include English, Arabic, German, French and Spanish. Monitor social media as well using our API and retrieve your data all with simple API calls.
Sentiment Analysis for Social Media - The multilingual sentiment analysis API (with exceptional accuracy, 83.4% as opposed to industry standard of 65.4%, and available in Mandarin) from Chatterbox classifies social media texts as positive or negative, with a free daily allowance to get you started. The system uses advanced statistical models (machine learning & NLP) trained on social data, meaning the detection can handle slang, common misspellings, emoticons, hashtags, etc.
Skyttle - Skyttle API is designed to turn any text into constituent terms (meaningful expressions), entities (names of people, place and things), and sentiment terms. Languages supported are English, Spanish, French, German, Chinese, Swedish, Greek, Czech, Italian and Russian.
Text-Processing - Sentiment analysis, stemming and lemmatization, part-of-speech tagging and chunking, phrase extraction and named entity recognition.
Stemmer - This API takes a paragraph and returns the text with each word stemmed using porter stemmer, snowball stemmer or UEA stemmer
SpringSense Meaning Recognition - The fastest and most accurate Meaning Recognition (Word Sense Disambiguation) API in the world. Recognises any nouns in a body of text and allows you to provide a rich user-interface with meaning definitions.
LanguageTool - Style and grammar checking / proofreading for more than 25 languages, including English, French, Polish, Spanish and German.
DuckDuckGo - DuckDuckGo Zero-click Info includes topic summaries, categories, disambiguation, official sites, !bang redirects, definitions and more. You can use this API for many things, e.g. define people, places, things, words and concepts; provides direct links to other services (via !bang syntax); list related topics; and gives official sites when available

You can also find our list of 50 Machine Learning APIs here and list of Sentiment Analysis APIs here.

25 Useful APIs for Book Discovery

Thu, 25 Apr 2013 14:09:00 -0400

We are proud to be part of the first Publishing Hackathon! The organizers of the Publishing Hackathon are inviting designers, engineers, programmers, an entrepreneurs to spend 36 hours together in teams to developer a new approach to digital book discovery.

The Publishing Hackathon will take place on May 18th and 19th at The Alley NYC, the leading digital co-working space in New York. The participants will be briefed by a cross-section of book publishing leaders, and then will form teams to create apps, websites, programming or businesses that can address the issue of book discovery in this rapidly evolving landscape. At the end of the weekend, a team of judges from publishing, technology, media and venture capital will identify the 3 to 5 most promising finalists from dozens of teams participating.

To help out developers in the Publishing Discovery hackathon, here are some APIs that could complement in solving the book discoverability problem:

Search:

Faroo Web Search - If you need an easy to use alternative to Google Web Search API (depreciated), Yahoo Boss (commercial) or Bing Web Search API
Mimvi App Search - Mimvi App Search provides mobile app search, discovery and recommendation services. Based on proprietary biomimetic algorithms, Mimvi App Search delivers the best coverage and most relevant mobile apps, mobile content and mobile products. Coverage includes: iTunes, Google Play (Android), Windows Phone 7, Windows Phone 8, Facebook, Blackberry, Web Apps.

Text Processing/Word Disambiguation:

Repustate Sentiment and Social Media Analytics - a sentiment analysis API that has “Nouns” and “Verbs” endpoints.
Spellcheck - fix spelling mistakes with this API.
SpringSense Meaning Recognition - The fastest and most accurate Meaning Recognition (Word Sense Disambiguation) API in the world. Recognises any nouns in a body of text and allows you to provide a rich user-interface with meaning definitions

Content Conversion:

OCR recognition service - Ocrapiservice.com is an cloud based optical recognition engine. We take images as input and we reply with text as output. (e.g. find text in surroundings and search relevant content)
SwissKnifeDocs - convert your documents into other document formats.
ConvertAPI - Several endpoints to convert content to PDF
Text2HTML - Converts text to HTML

Summarization:

Stremor Automated Summary and Abstract Generator - Use the Automated Summaries API to generate instant 350 character (+/- 10%) summaries of long content from text or URLs. (Like Summly!)
ML Analyzer - has a bunch of text processing endpoints, including an Article Summarizer.

Translation:

Translation Cloud - Simple to use API to get machine and human translation in over 50 languages. Powered by Translation Cloud’s network of 15,000 linguists.
Translation Memory - MyMemory is the world’s largest Translation Memory. It contains billions of words translated by professional translators. MyMemory will give you a machine translation (Google, Microsoft or our) only when a human translation is not available.

App Generator:

Apps Builder - AppsBuilder is a do it yourself platform for mobile applications development. No coding skills are needed! Trough a single building process users can create an app for iOS (iPad, iPhone/iPod), Android (Tablets & Phones) and Windows Phone.
MovixIt Mobile Apps - It enables you to create on the fly mobile applications that run on Android and Iphone.
Feed.nu - an api that generates an android application out from a RSS feed. Enter your own strings, colors, images and settings and then generate your own custom android application.

Promotion/Awareness:

Yupiq - A social media promotions platform to increase sales and brand awareness by rewarding customers for sharing content with their friends.

Text-To-Speech (TTS):

Text-to-speech - Instantly generates high quality speech from text.
Voice Text-to-Speech - allows conversion of textual content to speech
Text to Speech - A really simple api that lets you convert text to speech. It’s 100% free for unlimited usage.

Cross-vertical (e.g. find related books based on favorite song/lyrics):

Spotify Web - The Spotify Web API enables you to search and lookup metadata about artists, tracks and albums using a REST interface.
Musixmatch - The fastest, most powerful and legal way to display lyrics on your website or in your application. Today. A complete music catalogue featuring 640k artists and 13M of tracks organized by albums. A powerful search engine including all our library is available through our api. Every artist/track/album has a score (1-100) based on our api requests, this is our rating.
Pinterest (unofficial) - allows you to search for pins, and user-specific pins, likes, and boards.
Imgur - hosting for the most viral images on the web.

Prototyping:

Lorem Text Generator - Generate lorem ipsum filler text with this easy to use API.

Unicorn has become Unirest

Wed, 24 Apr 2013 16:35:00 -0400

Yesterday we released our newest open source project, which we called unicorn, unfortunately that was also the name of two existing open source projects. We really dropped the ball and have been working hard to clean up the mess. Our apologies to anyone in the Ruby or Python community who was confused or pissed off. Hopefully you can look past our mistake and see the value in this project.

Since yesterday we have decided on a new name, Unirest, which we think is a great fit for the project. We also made a new website, changed out all the docs, and updated our website which autogenerates unirest API snippets.

What problem does Unirest solve?

We’ve noticed that a lot of people have trouble wrapping their head around using public and private API’s. With API keys, varying endpoint architechture, differing response types, etc… there is a lot to think about. Unirest tries to simplify the process by abstracting a lot of the boilerplate and focusing on the core REST verbs that we all know and love (GET, POST, PUT, UPDATE, DELETE). Similarly its methods and response structure are the same in all the supported languages. It works for all REST APIs, available both on our own mashape API marketplace and abroad.

We truly believe that APIs are changing the way we make software and by extension the world at large. Never has so much power been in the hands of a solo developer with an idea. Our hope is that the developer community that we love will find this project helpful and build a ton of great stuff on top of it.

List of 16 Sentiment Analysis APIs

Wed, 24 Apr 2013 01:30:00 -0400

Just a few days back we posted a List of 40+ Machine Learning APIs.

The APIs below are a Sentiment Analysis subset group from that Machine Learning API list. Sentiment Analysis refers to “the application of natural language processing, computational linguistics, and text analytics to identify and extract subjective information in source materials.”

We hope you’ll it find useful!

TweetSentiments - Returns the sentiment of Tweets. Two online APIs call the Twitter API to analyze Tweets from a given Twitter user or Tweets returned by a Twitter search query. The offline API analyzes texts of Tweets you’ve already got, one Tweet at a time.
Sentiment Analysis for Social Media - The multilingual sentiment analysis API (with exceptional accuracy, 83.4% as opposed to industry standard of 65.4%, and available in Mandarin) from Chatterbox classifies social media texts as positive or negative, with a free daily allowance to get you started. The system uses advanced statistical models (machine learning & NLP) trained on social data, meaning the detection can handle slang, common misspellings, emoticons, hashtags, etc.
Repustate Sentiment and Social Media Analytics - Repustate’s sentiment analysis and social media analytics API allows you to extract key words and phrases and determine social media sentiment in one of many languages. These languages include English, Arabic, German, French and Spanish. Monitor social media as well using our API and retrieve your data all with simple API calls.
Chinese Sentiment Analysis for Social Media - 此API适用于中文社交媒体的情感分析（例如新浪微博），能针对每一条消息进行情感分类：正面或负面。该系统基于社交媒体，能够充分利用俚语，特殊词语等新新网络用语。请注意：该免费版本提供每天500条消息分类 - 超过此上限，将会被额外收费。
Sentiment Analysis Spanish - Sentiment analysis for Spanish language of any given tweet.
Viralheat Sentiment - Viralheat sentiment is free API and allows users to submit short chunks of text for sentiment scoring.
Text Processing - The WebKnox text processing API lets you process (natural) language texts. You can detect the text’s language, the quality of the writing, find entity mentions, tag part-of-speech, extract dates, extract locations, or determine the sentiment of the text.
ML Analyzer - Text Classification, Article Summarization, Sentiment Analysis, Stock symbol extraction, Person Names Extractor, Language Detection, Locations Extractor, Adult content Analyzer.
nlpTools - Text processing framework to analyse Natural Language. It is especially focused on text classification and sentiment analysis of online news media (general-purpose, multiple topics).
Chinese Analytics - Soshio allows companies to quickly expand their understanding of the Chinese market. Its Chinese Analytics API provides Chinese text analytics and sentiment analysis capabilities for businesses to create their own social monitoring dashboard.
Text-Processing - Sentiment analysis, stemming and lemmatization, part-of-speech tagging and chunking, phrase extraction and named entity recognition.
Skyttle - Skyttle API is designed to turn any text into constituent terms (meaningful expressions), entities (names of people, place and things), and sentiment terms. Languages supported are English, Spanish, French, German, Chinese, Swedish, Greek, Czech, Italian and Russian.
Truthy - Write scripts to work with our data, statistics, and images using the API. Download tweet volume over time, network layout, and statistics about memes and users, such as predicted political partisanship, sentiment score, language, and activity.
Speech2Topics - Yactraq Speech2Topics is a cloud service that converts audiovisual content into topic metadata via speech recognition & natural language processing. Customers use Yactraq metadata to target ads, build UX features like content search/discovery and mine Youtube videos for brand sentiment. In the past such services have been expensive and only used by large video publishers. The unique thing about Yactraq is we deliver our service at a price any product developer can afford.
Excitement Gauge for Social Media - This API is essential for measuring online audiences. Using advanced machine learning it automatically measures the excitement levels within social statuses. You’ll have heard of buzz (number of messages) but buzz is basic - it doesn’t tell you anything about the messages being posted. The Excitement Gauge allows you to measure quality on top of quantity. Combined with Chatterbox Sentiment Analysis this API is designed to support advertising revenues, sales promotions, product launches and loyalty programs across timelines, geographies and languages . This API provides a far more accurate measure of how successful, engaging and receptive your audience is across the digital landscape (online, TV and advertising).
Anger Detection for Social Media - This unique API will revolutionise your service levels, protect your brand and monitor both sales and promotional campaigns. Designed specifically for social media this API automatically measures the anger levels within social messages so you can quickly highlight action points. Combined with Chatterbox Sentiment Analysis, Anger Detection is designed to protect your brand and service interaction with an online audience.

You can also find our list of 40+ Machine Learning APIs here and List of Natural Language Processing (NLP) APIs here.

Do you know other Sentiment Analysis APIs? Send chris@mashape.com a message :)

Releasing Unicorn into the wild

Tue, 23 Apr 2013 16:46:00 -0400

UPDATE: We’ve got a lot of comments and suggestions on the ‘Unicorn’ naming. We apologize for the confusion and are working on finding a new name. Have a suggestion? Tweet us @mashape

UPDATE #2: We picked a new name, unirest.io is up and running.

UPDATE #3: Check our latest post on “Unicorn has become Unirest”

Here at mashape we use a lot of open source. Our tech stack includes a lot of popular frameworks and libraries which have allowed us to rapidly iterate and develop what we feel is the best cloud API proxy and marketplace the universe has ever seen. As part of our platform we have provided auto generated client libraries that allowed our users to get started with a mashape API right away.

About a month ago we decided to revamp the project and create Unicorn, an open source set of libraries that can consume not only API’s available on mashape, but any API you might encounter. And we built it in Java, PHP, Python, Ruby, and Objective-C with more languages planned for the future.

Enough about us, today is all about you. We truly believe that APIs are changing the way we make software and by extension the world at large. Never has so much power been in the hands of a solo developer with an idea. We have already built out a dedicated unicorn website complete with documentation and a very pretty unicorn (thanks @dshaw!) and the source is all hosted on GitHub. Our hope is that the developer community that we love will find this project helpful and build a ton of great stuff on top of it.

What problem does it solve?

We’ve noticed that a lot of people have trouble wrapping their head around using public and private API’s. With API keys, varying fundamental architechture, differing response types, etc… there is a lot to think about. Unicorn tries to simplify the process by abstracting a lot of the boilerplate and focusing on the core REST verbs that we all know and love (namely GET, POST, PUT, UPDATE, DELETE). Similarely it’s methods and response structure are the same in all the supported languages.

What’s next?

You can count on us to keep the project alive and updated, but our main goal isn’t maintaining a project, it’s helping developers build stuff. So go out and download unicorn and start consuming some APIs and mash them together into something that will change the world. Or your pocketbook.

Yoda teaches APIs

Mon, 22 Apr 2013 15:23:00 -0400

Into APIs now, Yoda is…

Consuming Machine Learning APIs with Python

Thu, 18 Apr 2013 21:39:00 -0400

In this short tutorial we will use Python in our Mac Terminal to consume one of the Machine Learning APIs in this list. Let’s pick DuckDuckGo.

DuckDuckGo lets us define people, places, things, words, and concepts. It also provides a list of related topics. This is very useful if you want to get some context on a certain text.

Although we’re picking DuckDuckGo, note that the steps below also applies to the rest of the APIs in the Machine Learning APIs list, and all Mashape APIs. That means you can swap in and swap out APIs for your applications easily.

Let’s get started!

Sign up for an account in Mashape here. Mashape is an API Marketplace where you can find new and useful APIs for your application. To access these APIs, you would need a Mashape account.
Once you have signed up, login to Mashape and go to DuckDuckGo’s interactive API documentation here. You can try the test console by keying in a word (e.g. “Mashape”) on the “q” parameter, and hit “Test Endpoint”.

The test console gives you a way to “preview” the response of the API endpoints before you actually code.

You should see a response similar to below:
Now let’s get to the Python part. On the test console, you are given options to pick a client library of your preferred language. Let’s click on the Python link. It will show us the snippet of code to call this endpoint.

The code snippet uses an open source library called Unirest. Without going into too much detail, Unirest makes it dead easy to make HTTP requests for a variety of programming languages. We will need to get this Unirest library to make an HTTP call in Python.
We’ll get back to that Python code snippet in a bit. Let’s go to the link shown above to download the Unirest library for Python. http://unirest.io/#python. It shows us how easy it is to install the library:

(Note: ”pip” is a utility to install Python packages. You can install it by following the instructions here.)
We will type in the commands above in our Terminal. Here’s what our Terminal looks like after successfully installing Unicorn.

That’s not too hard isn’t it :)
The last step is to go to Python and import the Unicorn library and call the endpoint. These are mentioned in Steps 4 and 3 respectively. We’re going to use command-line Python below.

The whole chunk of response output in the picture above is the same as in Step 2. From here you can refer to the specific object attributes that you will use for the rest of your application.

And that’s it! To recap:

- We signed up for a Mashape account and tried the API endpoint using the Mashape test console

- We installed the Unirest library to make it easy for us to call the endpoint in Python

- We imported Unirest into our Python script and ran the snippet specified in the Mashape test console to get the output.

It is important to keep in mind that these concepts extend to other APIs in Mashape (not just the Machine Learning ones) and to other programming languages.

We hope this post was useful!

The Growth Hacking Manifesto

Wed, 17 Apr 2013 13:43:00 -0400

Our very own Montana Flynn has released a new project codenamed the growth hacking manifesto. If you haven’t heard the buzzword yet, you’re probably not living in San Francisco. The project is a complilation of his own notes, experiments, and thoughts on all things user acquisition and retention. If you have a company, product, API, or are remotely interested in tech startups we suggest you take a look.

In the first article, What The FSCK is Growth Hacking?, Montana explains what a growth hacker is and how they change the way we think of user growth. Here’s the TL;DR: A growth hacker is a cross between a developer and marketer.

We hope you enjoy the series, of course if you don’t or think that it’s missing some special sauce please let us know (or you can just yell at @montanaflynn directly on twitter).

List of 40+ Machine Learning APIs

Mon, 15 Apr 2013 19:11:00 -0400

Wikipedia defines Machine Learning as “a branch of artificial intelligence that deals with the construction and study of systems that can learn from data.”

Below is a compilation of APIs that have benefited from Machine Learning in one way or another, we truly are living in the future so strap into your rocketship and prepare for blastoff.

OCR recognition service - Ocrapiservice.com is an cloud based optical recognition engine. We take images as input and we reply with text as output. Checkout http://ocrapiservice.com to learn more.

Face (Lambda) - A computer vision api for facial recognition and facial detection that is a perfect face.com replacement. We currently have a free api for face detection.

Viralheat Sentiment - Viralheat sentiment is free API and allows users to submit short chunks of text for sentiment scoring.

Chinese Analytics - Soshio allows companies to quickly expand their understanding of the Chinese market. Its Chinese Analytics API provides Chinese text analytics and sentiment analysis capabilities for businesses to create their own social monitoring dashboard.

MyMemory - Translation Memory - Get a better translation! MyMemory is the world’s largest Translation Memory. It contains billions of words translated by professional translators. MyMemory will give you a machine translation (Google, Microsoft or our) only when a human translation is not available.

Text-Processing - Sentiment analysis, stemming and lemmatization, part-of-speech tagging and chunking, phrase extraction and named entity recognition.

Skyttle - Skyttle API is designed to turn any text into constituent terms (meaningful expressions), entities (names of people, place and things), and sentiment terms. Languages supported are English, Spanish, French, German, Chinese, Swedish, Greek, Czech, Italian and Russian.

Face and Scene recognition provided by ReKognition.com - Face.com alternative! Our fast, robust and scalable rekognition engine can do facial detection, crawling, recognition, scene understanding! It can be automatically trained using images and tags on Facebook! Please visit http://rekognition.com/ or email me in mashape to register free quota!

Query Classification - This API can be used to find the topic of a keyword query. The topic for the query will be chosen from over 1500 predefined topics. This works best with query phrases like - “Star Wars”, “Dell Inspiron 1420 price” This technology has been provided by Query Dynamics LLC (http://www.querydynamics.com). This technology is being used to categorize tweets at a website called TweetDynamics (http://www.tweetdynamics.com/) To get the class for the query “Star Wars” submit a GET request to - http://tweetdynamics.com/perl/parseQueryJ?query=Star%20Wars The following JSON code will be returned: {“Answer”: sci-fi movie} The category of the query is within the by the Answer tag. Some of the uses of this api are to create tools for online query analysis and offline query log analysis.

Face Recognition - Stephen here from Lambda Labs. For sample code and a graphical demo, check out http://api.lambdal.com/docs. Our API provides face recognition, facial detection, eye position, nose position, mouth position, and gender classification. If you have any questions ask! Just send an email to s@lambdal.com, or call (802) 598-6343. Happy Hacking! -Stephen

Jeannie - Jeannie (Voice Actions) is a virtual assistant with over two Million downloads, now also available via API. The objective of this service is to provide you and your robot with the smartest answer to any natural language question, just like Siri. This service provides an interface to the standard functions that users demand of modern voice assistants. For example chatting, looking up information, creating messages and much much more. It also provides useful metadata such as sentence analysis and entity extraction that goes beyond simple chatting and voice commands. Over 2 million users have already been in contact with this API: http://www.voice-actions.com/ Examples: https://weannie.pannous.com/api?input=hi&login=test-user https://ask.pannous.com/?input=hi Complete documentation: https://docs.google.com/document/d/1dVG_B5Sc2x-fi1pN6iJJjfF1bJY6KEFzUqjOb8NsntI/edit

Porn Filter - Scan images and image URLs to determine if they contain inappropriate content. Send us an image, and we will tell you if it’s likely to be porn. A note about scanning for porn: Nothing in this world is perfect. Our API looks at skin tones, shapes and other cues to identify images that are likely to be inappropriate, but nothing is ever going to be perfect. Just like virus scanning, spam filters and other things, some false positives and missed hits will occur.

Portrait3D - Portrait3D API is based on Avatarion’s Tethys 3D™ solution, and provides software developers with a technology to create animated facial models based on photos.

SpringSense Meaning Recognition - Free plan available! The fastest and most accurate Meaning Recognition (Word Sense Disambiguation) API in the world. Recognises any nouns in a body of text and allows you to provide a rich user-interface with meaning definitions. More exhaustive (and useful) bindings are available at: http://github.com/SpringSense/ruby-api-bindings http://github.com/SpringSense/java-api-bindings http://github.com/SpringSense/python-api-bindings

Idilia Sense Analytics - Precisely annotate text with fine senses using the world’s only API that disambiguates both common words (all parts of speech) and proper nouns (NEs) with near human accuracy. Use specialized recipes for well-formed text, queries, and social media (e.g. tweets). Get lexical annotation, statistical confidence scores, external links (wikipedia, twitter verified accounts, etc), and precise classification of NEs. Tags: disambiguation, wsd, text analytics, language, sense annotation, semantic, extraction ** For more documentation see at http://idilia.com/docs/rest_api/text-disambiguate ** Developer forum at http://groups.google.com/forum/?fromgroups#!forum/idilia-developers

ImageVision - NuditySearch - ImageVision’s NuditySearch - Recognizing nudity is a highly complex problem. NuditySearch tackles this problem by recognizing anatomical attributes and determining if there is nudity in images. Beta version of ImageVision’s NuditySearch Engine Find out more at www.ImageVision.com Keywords: nude, nudity, lewd, suspicious, improper, protect, anatomy, anatomical, image analysis, image detection, feature extraction, image scan.

SkyBiometry Face Detection and Recognition - An easy to use Face Detection and Recognition API. You must have an application created in your account at SkyBiometry to use it (sign up at https://www.skybiometry.com/Account/Register if you don’t have account yet).

Question-Answering - The WebKnox question-answering API allows you to find answers to natural language questions. These questions can be factual such as “What is the capital of Australia” or more complex.

TweetSentiments - Returns the sentiment of Tweets. Two online APIs call the Twitter API to analyze Tweets from a given Twitter user or Tweets returned by a Twitter search query. The offline API analyzes texts of Tweets you’ve already got, one Tweet at a time.

Bypass Captcha - Bypass any captchas you meet and free your hands.

Text Processing - The WebKnox text processing API lets you process (natural) language texts. You can detect the text’s language, the quality of the writing, find entity mentions, tag part-of-speech, extract dates, extract locations, or determine the sentiment of the text.

Infatics Face Detection - Simple face detection API.

Sentiment Analysis for Social Media - The multilingual sentiment analysis API (with exceptional accuracy, 83.4% as opposed to industry standard of 65.4%, and available in Mandarin) from Chatterbox classifies social media texts as positive or negative, with a free daily allowance to get you started. The system uses advanced statistical models (machine learning & NLP) trained on social data, meaning the detection can handle slang, common misspellings, emoticons, hashtags, etc. You can use this API for free. If you require more flexibility and larger scale you can move up to a paid-for plan. Full details on the awesomeness of the API can be found at chatterbox.co/api. See a CASE STUDY of this API in use here: http://bit.ly/Wlku1z. Please note: The free plan offers 500 statuses classified per day - any use over this will be charged for. If you are using more than 500 statuses per day it is more cost effective to upgrade to one of the low cost, paid-for plans or contact us direct for an Enterprise offering.

ColorTag - ColorTag is a powerful API for color detection. Taking an image file (or URL) as input ColorTag produces a list of text labels and hex RGB values that can be then used as tags for a certain image or item. The API can sort tags by relevance (detecting colors of objects on the image, e.g. a dress, a car, etc.) or simply by weight in the image. First mode is perfect for e-commerce applications, allowing to automatically tag items with colors by simply providing a photo or a thumbnail and build a color tag cloud, so users are able to search items by color (e.g. red). Weight sorting mode can be useful for photos, wallpapers or other images without well-defined objects on them to analyze the palette in general. Text color labels can be assigned with different precision (just basic colors, W3C-compatible colors, precise colors, etc.).

Chinese Sentiment Analysis for Social Media - This API performs Chinese Sentiment Analysis for Social Media (such as Sina Weibo statuses). For a provided message it classifies the status as positive or negative. The system is designed specifically for social media and so it can take advantage of slang and niche language. This API is in testing mode and should be considered a BETA release. The API is likely to change at any point - it is unstable and the results should only be used for testing. Please note: The free plan offers 500 statuses classified per day - any use over this will be charged for. 此API适用于中文社交媒体的情感分析（例如新浪微博），能针对每一条消息进行情感分类：正面或负面。该系统基于社交媒体，能够充分利用俚语，特殊词语等新新网络用语。请注意：该免费版本提供每天500条消息分类 - 超过此上限，将会被额外收费。

Nudity Detection Service - This API allows developers to check for nudity in images.

Animetrics Face Recognition - he Animetrics Face Recognition API can be used to detect human faces in pictures. Information on facial features or “landmarks” is returned as coordinates on the image.. Animetrics Face Recognition will also detect and return the orientation, or “pose” of faces along 3 axes. We plan to add functionality for facial matching very soon! A special capability called “SetPose” is also available. SetPose images are frontal view mugshots of a face which have been zero-corrected for pitch (x axis), yaw (y axis), and roll (z axis). As long as within the input facial image both eyes are visible, then SetPose will pose correct that facial image to x=y=z=0, a perfect frontal.

Stemmer - This API takes a paragraph and returns the text with each word stemmed using porter stemmer, snowball stemmer or UEA stemmer

Term Analysis - Given a text, this API returns the lemmatized text

Truthy - Write scripts to work with our data, statistics, and images using the API. Download tweet volume over time, network layout, and statistics about memes and users, such as predicted political partisanship, sentiment score, language, and activity.

Semantic Analytics - Text analysis API including wordnet synsets,relation extraction,named entity recognition and classification,lemmatization,part of speech tagging,tokenization, and semantic role labeling. Also includes wikipedia types from dbpedia, YAGO ( so it knows about artists, presidents, ..) as well as disambiguation using wikipedia redirects (President Abraham Lincoln redirects to Abraham Lincoln on Wikipedia) useful for detecting who a person might be.

Anger Detection for Social Media - This unique API will revolutionise your service levels, protect your brand and monitor both sales and promotional campaigns. Designed specifically for social media this API automatically measures the anger levels within social messages so you can quickly highlight action points. Combined with Chatterbox Sentiment Analysis, Anger Detection is designed to protect your brand and service interaction with an online audience. Identifying badvocates (unhappy clients) will ensure your product launch runs smoothly, your campaign is on track and that no viral negative influence impacts sales revenues. This API has proven successful with large volumes of data, at the individual message level and offers ease of integration with an enterprise system for immediate attention within the relevant area of your business. Historical and current measurement will also be an attractive reason of why to buy this API.

Excitement Gauge for Social Media - This API is essential for measuring online audiences. Using advanced machine learning it automatically measures the excitement levels within social statuses. You’ll have heard of buzz (number of messages) but buzz is basic - it doesn’t tell you anything about the messages being posted. The Excitement Gauge allows you to measure quality on top of quantity. Combined with Chatterbox Sentiment Analysis this API is designed to support advertising revenues, sales promotions, product launches and loyalty programs across timelines, geographies and languages . This API provides a far more accurate measure of how successful, engaging and receptive your audience is across the digital landscape (online, TV and advertising). Historical and current measurement will also be an attractive reason of why to buy this API.

Ping.it - Send fast and direct tips about good content to any of your friends, co-workers and family.

DocumentParser - Extracts text and metadata from a variety of formats (HTML, Office, PDF, XML, etc…) and identifies the mime-type of a document. The number and nature of the metadata returned depends on the mime-type and content of the document. All documents will have at least a ‘Content-Type’ key/value and often a ‘title’. Based on your level of subscription, the API can also detect the language of a document. The identification is based on statistical methods and has over 99% precision for 53 languages.

Diffbot - Diffbot extracts data from web pages automatically and returns structured JSON. For example, our Article API returns an article’s title, author, date and full-text. Use the web as your database! We use computer vision, machine learning and natural language processing to add structure to just about any web page.

Sentiment Analysis Spanish - Sentiment analysis for Spanish language of any given tweet.

PeerReach - The PeerReach API allows you to give context to content produced by users. Currently we only support Twitter users but will accept other networks in the near future.

nlpTools - Text processing framework to analyse Natural Language. It is especially focused on text classification and sentiment analysis of online news media (general-purpose, multiple topics).

ML Analyzer - Analyze a block of text for adult content, return true/false

Repustate Sentiment and Social Media Analytics - Repustate’s sentiment analysis and social media analytics API allows you to extract key words and phrases and determine social media sentiment in one of many languages. These languages include English, Arabic, German, French and Spanish. Monitor social media as well using our API and retrieve your data all with simple API calls.

Stremor Automated Summary and Abstract Generator - Use the Automated Summaries API to generate instant 350 character (+/- 10%) summaries of long content from text or URLs. Summaries are returned as highly readable paragraphs with complete sentences for the best end-user experience.

AI Netizens: The State of Agents Online

Mon, 08 Apr 2013 20:22:00 -0400

Watch Jeff Kramer’s (@jeffk) talk on the… “history of agents and bots, and how we design and develop bot platforms for the future. We touch on MUD and IRC bots, Weavrs, Siri, Google Now, the Internet of Things, PaaS architectures, the API of Me, Xenobi and a bunch of other things”

On Mashape, Jeff makes an interesting point on the purpose of an API Marketplace to facilitate the transferability of processing units through APIs. He uses the Chatterbox Sentiment Analysis API as an example.

You can watch the talk below (and skip to the Mashape part here).

Congratulations to Ronaldo Barbachano for winning our Mashape...

Sun, 07 Apr 2013 15:59:00 -0400

Congratulations to Ronaldo Barbachano for winning our Mashape April API Hack! He used the Speech2Topic API from Yactraq and created a visualization tool to show topics as they occur on a video while it’s playing. He won a Lego Mindstorm NXT 2.0!

You can try his application here and here.

Until our next hackathon!

APIs based on Machine Learning

Fri, 05 Apr 2013 17:24:00 -0400

Here are some Mashape APIs based on Machine Learning to prepare you for our April Hackathon tomorrow (April 6th).

Example of Face Recognition using Javascript:

http://blog.mashape.com/post/45712257463/face-recognition-using-javascript-and-mashape

Example of Yactraq Video Topic Extraction using Javascript:

http://blog.mashape.com/post/47058276181/video-topic-extraction-and-transcription-tutorial

Nodejs wrapper for Yactraq

https://github.com/redcap3000/yactraq-nodejs-mashape/

Creating social “Signals” with Ping.It

https://ping.it/Develop-Ping.it-Signals-on-Mashape-Platform-Tutorial.htm

Example of a Face Recognition login page

http://fadomire.futureauth.jit.su/

Video topic extraction and transcription

Wed, 03 Apr 2013 19:04:00 -0400

This is a quick tutorial on how to transcribe and extract topics from videos using the Yactraq Speech2Topic, API in Mashape.

(Note: Yactraq is giving away $500 to the winner Yactraq category for this coming Mashape April Hackathon Apr 6th, 2013. We require a number of teams to submit to this category to unlock the prize :))

This is particularly useful if you’re trying to get context in the video, which can be used to query ad services to pull relevant content.

The process is very simple, we send a link (YouTube) of a video to be processed to the Yactraq API, and we get back a result with information on topics and transcribed words. This result metadata will indicate when in the video these topics and words occur.

Here’s a sample of the response from a processed video:

https://gist.github.com/ismaelc/5305880

And here’s what an app that consumes it looks like in action:

Let’s get started!

Step 1: Get a Mashape key. All calls to any Mashape API requires a Mashape key. If you haven’t done so yet, you can sign up for an account at http://www.mashape.com/signup. You can get your Mashape key in your Dashboard.

Step 2: Request for a Yactraq developer account. We need the Yactraq account so we can make a proper request to process our videos. To request for an account, follow the instructions here.

Step 3: Submit a video processing request through the Yactraq Speech2Topics API. To do this, you would need the YouTube url of the video you want to extract topics from. Make a note of the video id of it as well (e.g. http://www.youtube.com/watch?v=zFOnOLXrMBQ, the video id is “zFOnOLXrMBQ”). We also need information from our Yactraq account, particularly “secret” and “adset” which you can get by logging in to http://api.yactraq.com/ . Also set “start” and “tx” to 1.

My request looks like below:

Then hit “Test Endpoint”. If your request is successful, you would get a response like below:

Step 4: Get some coffee and wait for your request to be processed (According to Yactraq the waiting time takes about half or a third of your video length).

Then you can try to check the status by calling the same endpoint, only this time set “start” to “0” or leave it blank. This will tell the API to get the results, instead of initiating a new request.

If your request is successful, you should get a lengthy response like the one I showed you at the beginning of this tutorial. What you didn’t see there are the transcription results which look like this:

https://gist.github.com/ismaelc/5306111

For both parts of the response, there are “ts” values per topic or word transcribed. You can use this ts (timestamp in seconds) value to see where the words and topics occurred in the video.

I have created a sample app that shows how this can be done here. (the app in the video at the top).

We hope this will get you started in building your own apps that use the Speech2Topic API in Mashape. Let us know if you have suggestions or comments for our next tutorial!

Introducing the ping status report

Tue, 02 Apr 2013 15:47:00 -0400

We’re rolling out a new detailed status report for API Providers. It gives you a second-by-second report on your API health, as pinged from the Mashape proxy.

The goal is to give API providers a high-fidelity report on how their API is doing at minute intervals.

You may remember sometime back, we also rolled out the Status Snapshot Display for APIs. This a yet another continuation of our commitment to provide the best API report.

Currently, it can only be accessed through the email notification you get when your API is down. We would love to hear your feedback on where we can place this report in Mashape. Send us a note at support@mashape,com