IBM Watson cognitive services one pager

Jeronimo De Leon

With the Watson Hackathon coming up on May 4–5 I thought it would be useful to have a one pager of the current 13 cognitive services available.

  • Concept Expansion
  • Concept Insights
  • Language Identification
  • Machine Translation
  • Message Resonance
  • Personality Insights
  • Question and Answer
  • Relationship Extraction
  • Speech to Text
  • Text to Speech
  • Tradeoff Analytics
  • Visual Recognition
  • Visualization Rendering

IBM also recently acquired AlchemyAPI significantly expanding Watson’s ecosystem. The AlchemyAPI has two core services

  • AlchemyLanguage
  • AlchemyVision

The Watson Developer Cloud Github account has a collection of sample and starter apps that use the Watson REST APIs and SDKs.

Certain Watson services are contextually specific and knowledgeable depending on the domain model and content set they are connected to. Most of Watson services are still in Beta and some only contain a specific data set / domain to work from. I am guessing for the Hackathon that all the services will be opened up but below gives a snapshot of what those services are and their data sets / domains.

Concept Expansion

This service analyzes text and interprets its meaning based on usage in other similar contexts.

Example Usage
Concept Expansion could identify that ‘The Big Apple’ refers to New York City and that ‘getting in touch’ means communicating by email, letter or phone.

What it does
You can input: The starting point word, a few terms that are examples of that word, and a data set to analyze

And the service will output: A ranked list of terms with contextual similarity to the starting point word based on analysis of the data set

Beta-data set
This service analyzes input text against a choice of two pre-defined data sets:

  • Periodically updated random tweets
  • Medical transcript samples from MTSamples

Further documentation
Demo

Concept Insights

This service maps user-input words to the underlying concepts of those words based on training on English Wikipedia data.

Example Usage
Some examples of Concept Insights service use cases are for an application like an expertise locator — for example, when trying to locate an expert on ‘cognitive systems’, the Concept Insights service can help find people who work on related areas such as ‘neural networks’ — or to drive increased customer engagement on a customer-facing site — for example, when searching for news about ‘natural disasters’, the Concept Insights service can retrieve content referring to ‘hurricanes’ or ‘remote sensing’.

What it does
You can input: Your own content and queries. Text (Unicode) and HTML for the content. For queries, an API for entering concepts in an auto-complete box is also provided.

And the service will output: A list of content that is relevant to your input, together with useful information for visualizing it (most important concepts, information on how to render the text emphasizing the concepts, and so on). Can also identify concepts that are related to other concepts.

Beta-data set:

  • English Wikipedia, used as a concept graph
  • Your own data set

Further documentation
Demo

Language Identification

This service detects the language in which text is written. This helps inform next steps such as translation, voice to text, or direct analysis.

Example Usage
Language Identification service enable machines to recognize that the phrase ‘Wie geht es Ihnen’ is written in German.

What it does
You can input: Plain text in UTF-8 encoding

And the service will output: 5-letter ISO language code

Beta — domains
Today, the service can identify many languages: Arabic; Chinese (Simplified); Chinese (Traditional); Cyrillic; Danish; Dutch; English; Farsi; Finnish; French; German; Greek; Hebrew; Hindi; Icelandic; Italian; Japanese; Korean; Norwegian (Bokmal); Norwegian (Nynorsk); Portuguese; Spanish; Swedish; Turkish; Urdu.

Further documentation
Demo

Machine Translation

Translate text from one language to another

Example Usage
A French speaking help desk representative is assisting a Spanish speaking customer through a chat session and is able to interact through the translation service.

What it does
You can input: Plain text input in the selected input language

And the service will output: Translated text in the selected output language.

Beta-domain
Translation is available among English, Brazilian Portuguese, Spanish, French and Arabic.

Further documentation
Demo

Message Resonance

This service analyzes draft content and scores how well it is likely to be received by a specific target audience. This analysis is based on content that has been written by the target audience itself, such as fans of a specific sports team or new parents.

Example Usage
Among people active in cloud computing discussions, option A content is likely to resonate very well, option B poorly, and option C moderately well.

What it does
You can input: Term to evaluate and the audience/community to measure against (e.g. — the Cloud community or Big Data & Analytics community)

And the service will output: A score ranking how well the term will be received by the intended audience/community. Higher resonance means the message has a higher likelihood of being viewed and shared.

Beta — data set and domain:
This service analyzes input text against a choice of two pre-defined data sets of ~1M tweets updated every few hours covering:

  • Cloud
  • Big Data

Further documentation
Demo

Personality Insights

Enables deeper understanding of people’s personality characteristics, needs, and values to help engage users on their own terms. Using linguistic analytics to infer cognitive and social characteristics, including Big Five, Values, and Needs, from communications that the user makes available, such as email, text messages, tweets, forum posts, and more. By deriving cognitive and social preferences, the service helps users to understand, connect to, and communicate with other people on a more personalized level.

Example Usage
The service can analyze text based on a customer’s twitter stream to help a travel agency decide between leading with a budget or luxury trip offer.

What it does
You can input: JSON, or Text or HTML (such as social media, emails, blogs, or other communication) written by one individual

And the service will output: A tree of cognitive and social characteristics in JSON or CSV format

Beta — data set:
This service uses any content.
Minimum 1000 words of text written by one individual; 2000 or more words recommended.

Further documentation
Demo

Question and Answer

Interprets and answers user questions directly based on primary data sources (brochures, web pages, manuals, records, etc.) that have been selected and gathered into a body of data or ‘corpus’. The service returns candidate responses with associated confidence levels and links to supporting evidence.

Example Usage
Healthcare: What is a stroke? What is the cause of Wilson Disease? Travel: Where is the best place to stay in Prague?

What it does
You can input: User questions

And the service will output: List of candidate passages, Confidence scores, Links to supporting evidence from the corpus

Beta — data set:
The beta service is optimized to use with two pre-trained data sets:

  • Healthcare data (including Healthfinder and CDC Health Topics)
  • Travel data (including Wikivoyage, TSA, and CDC Travel)

Further documentation
Demo

Relationship Extraction

From unstructured text, Relationship Extraction can extract entities (such as people, locations, organizations, events), and the relationships between these entities (such as person employed-by organization, person resides-in location).

Example Usage
Relationship Extraction can analyze a news article and, based on statistical modeling, the service will perform linguistic analysis of the input text, find spans of text that refers to entities, cluster them together to form entities, and extract the relationships between the entities. Note: Relationship Extraction models are domain-specific and work best with in-domain input.

What it does
You can input: Text news articles

And the service will output: Entities from the text and relationships between those entities in xml format

Beta - domain:
The service is optimized for news articles or other news-related text

Further documentation
Demo

Speech to Text

This service converts the human voice into the written word. This easy-to-use service uses machine intelligence to combine information about grammar and language structure with knowledge of the composition of the audio signal to generate a more accurate transcription.

Example Usage
The Speech to Text service can be used for anything in which text is the desirable service output, including voice control of applications, embedded devices, and vehicle accessories, transcription of meetings and conference calls, dictation of email and notes, and integration into many systems across various domains.

What it does
You can input: streamed or recorded audio

And the service will output: text transcriptions of the recognized words

Beta-data set
Intelligible English speech

Further documentation
Demo

Text to Speech

This service understands text and natural language to generate synthesized audio output complete with appropriate cadence and intonation.

Example Usage
Input any English text to generate speech output for assistance tools for the vision-impaired, reading-based education tools, and for multiple mobile applications.

What it does
You can input: English text

And the service will output: Synthesized audio based on the input text

Beta-data set
English or Spanish text

Further documentation
Demo

Tradeoff Analytics

This service helps people optimize their decisions while striking a balance between multiple, often conflicting, objectives. The service can be used to help make complex decisions like what mortgage to take or which laptop to purchase.

Example Usage
Tradeoff Analytics could help bank analysts or wealth managers to select the best investment strategy based on performance attributes, risk, and cost. It could help consumers purchase the product that best matches their preferences based on attributes like features, price, or warranties. It could also help physicians find the most suitable treatment based on multiple criteria such as success rate, effectiveness, or adverse effects.

What it does

You can input: A decision dilemma (e.g. What is the best car where my goals are Type, Price, and Fuel economy?)

And the service will output: JSON objects representing the optimal options, explanation for excluded options, highlight of trade-off between options AND representation of the optimal options in an HTML (graphical representation) to enable better exploration and selection

Further documentation
Demo

Visual Recognition

This service enables you to analyze the visual appearance of images or video frames to understand what is happening.

Example Usage
If you have a large collection of digital photographs, the Visual Recognition service could be used to organize them into albums based on visual themes such as cars, sports, or photos showing people. Similarly, if you have multiple groups of photos from different people, you could build an app to associate them semantically — for example, person A and person X have many baseball and beach pictures, while persons J, K, and L have many indoor photos.

What it does
You can input: JPEG images

And the service will output: a set of labels and likelihood scores (such as ‘soccer, 0.7’,’baseball, 0.3' in JSON)

Beta- domain:
There is a fixed set of visual models, which is a subset of over 1000 that we have created.

Further documentation
Demo

Visualization Rendering

Graphical representations of data analysis for easier understanding.
Render interactive data visualizations that can range from common business charts to more advanced layouts.

Example Usage
The service could represent neighborhood demographic data as mini pie charts showing income levels centered on geographic locations on maps, or as tree maps that can switch from looking at income by age to house size or by education level.

What it does
You can input: Data to be visualized and a visualization description

And the service will output: Interactive visualizations of the input data in your application

This service is an SDK that can be used to visualize any numeric data.

AlchemyLanguage

AlchemyAPI offers 12 API functions as part of its text analysis service, each of which uses sophisticated natural language processing techniques to analyze your content and add high-level semantic information. Browse the links below to learn more about each of the functions of AlchemyAPI’s text analysis service.

Text Analysis Functions:

Further documentation
Demo

AlchemyVision

AlchemyVision employs deep learning innovations to understand a picture’s content and context. Organizations across a variety of industries — ranging from publishing and advertising to eCommerce and enterprise search — can effectively integrate images as part of big data analytics being used to make critical business decisions.

AlchemyVision sees complex visual scenes in their entirety — without needing any textual clues — leveraging a holistic approach to understanding the multiple objects and surroundings in common smartphone photos and online images.

Further documentation
Demo

— —

If you are new to Watson here is a great summary video

More information on Watson can be found here

    Jeronimo De Leon

    Written by

    Product Marketing at www.Spell.run Deep Learning Platform / Founder of www.Welcome.AI

    Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
    Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
    Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade