Artificial Intelligence APIs: An Intro for Developers Who Must Build Smarter Applications

Post on 29-Nov-2014

2.439 views 3 download

description

If you’ve used applications like Siri, Pandora or Netflix, you’ve seen how artificial intelligence (AI) improves your experience with personalized recommendations and smart assistants. Now, learn how to put AI to work in your applications. During this webinar, Devin Harper, AI Research Scientist at AlchemyAPI, shares examples of how organizations are applying the newest AI technologies to gather and trend customer feedback, bucket and route inbound communications, build recommendation engines and more. Hear an overview of AI solutions ranging from image and text understanding to machine learning and prediction APIs and get tips on how you can get started today. In this session, you'll learn: - How AI helps developers create applications and services that process unstructured data [in the form of Tweets, emails, chats, articles, etc.] and turn it into actionable information - The kinds of easy-to-use, cloud technology available today - How to construct a simple AI-driven application relevant to your business (email analysis demo)

Transcript of Artificial Intelligence APIs: An Intro for Developers Who Must Build Smarter Applications

Devin Harper@phdharp

Artificial Intelligence APIs: For Developers Who Must Build Smarter Applications

Pioneer of web services for real-time text and image analysis

• Founded in 2005 • 40,000+ users • Used in 36+ countries • More than 3+ billion API calls monthly • Deep learning experts

Devin Harper, AI Research Scientist AlchemyVision

Meet the Expert

#deeplearningseries

What You’ll Learn Today

#deeplearningseries

1

2

3

#deeplearningseries

What is Artificial Intelligence (AI)?

We’re not talking about killer robots…

#deeplearningseries

Speech Recognition Machine Translation

Computer Vision Natural Language Processing

Question Answering Machine Learning #deeplearningseries

The AI Revolution…

#deeplearningseries

How are these methods being used?

• Driving recommendation engines

• Statisticians derive structured data from ambiguous information

• E-commerce

• Advertising & social media marketing

• Google Dataflow

• Consumer Products

#deeplearningseries

#deeplearningseries

#deeplearningseries

What APIs are available?

AT&T provides a straightforward API to convert speech data to natural text.

https://developer.att.com/apis/speech "

Speech Recognition

AI APIs

#deeplearningseries

curl "https://api.att.com/speech/v3/speechToText" \!

--header "Authorization: Bearer {ACCESS_TOKEN}" \!

--header "Accept: application/json" \!

--header "Content-Type: audio/amr" \!

--data-binary @{PATH_TO_FILE} \!

--request POST

APP IDEA Medical personnel

transcribe dictation of pt. notes into text

{! "Recognition": {! "Status": "Ok",! "ResponseId": "3125ae74122628f44d265c231f8fc926",! "NBest": [! {! "Hypothesis": "patient experiences chronic anxiety

! ! ! ! ! ! ! reaction to external stimulus",! "LanguageId": "en-us",! "Confidence": 0.9,! "Grade": "accept",! "ResultText": "patient experiences chronic anxiety

! ! ! ! ! ! ! reaction to external stimulus",! "Words": [! “patient", “experiences", “chronic”, “anxiety",! “reaction”, “to”, “external", "stimulus"! ],! "WordScores": [! 0.92, 0.73, 0.81, 0.96, 0.74, 0.97, 0.81, 0.69! ]! }! ]! }!}

AlchemyAPI provides an API for generating a list of keywords based on an input image.

http://www.alchemyapi.com/products/alchemyvision/ "

Computer Vision

AI APIs

#deeplearningseries

wget -O tags --post-file image.jpg “http://access.alchemyapi.com/calls/image/ ImageGetRankedImageKeywords?apikey={API_KEY}&outputMode=json”!

APP IDEA Ecommerce -

consumer takes photo & receives

product recommendation

{! "status": "OK",! "usage": "By accessing AlchemyAPI or using information generated …,! "totalTransactions": "4",! "imageKeywords": [ {! "text": "iphone",! "score": "0.99593"! },! {! "text": "apple",! "score": "0.622459"! }! ]!}

Computer Vision

AI APIs

#deeplearningseries

Enables computers to derive meaning from human or natural language input (i.e. text, articles, chats, emails, etc.)

Natural Language Processing (NLP) AI APIs

#deeplearningseries

NLP determines… WHO, WHAT, WHEN, WHERE and WHY …in unstructured data.

Use it to: •  Drive content recommendation

engines •  Track sentiment on Twitter •  Monitor social media for trends and

consumer preferences •  Determine how people feel about a

your brand or product •  Tag documents by topic •  Perform contextual ad targeting

APP IDEA Stock tracker – Follow news & consumer

perception re: companies you care

about

Natural Language Processing

AI APIs

#deeplearningseries

AlchemyAPI provides an extensive suite of NLP capabilities through REST APIs.

wget -O sentiment.txt “http://access.alchemyapi.com/calls/text/!

TextGetRankedNamedEntities?apikey={API_KEY}&outputMode=json&sentiment=1

&text=Google strengthens hardware expertise by acquiring Gecko Design”

wget -O entities.txt “http://access.alchemyapi.com/calls/url/

URLGetRankedNamedEntities?apikey={API_KEY}&outputMode=json&sentiment=1

&url=http://seekingalpha.com/news/1949845-google-strengthens-hardware-expertise-

by-acquiring-gecko-design”

Raw Text

Crawling a URL

Natural Language Processing

AI APIs

#deeplearningseries

Try the demo: http://www.alchemyapi.com/products/demo/alchemylanguage/ "

{! "text": "Google strengthens hardware expertise by acquiring Gecko Design",! "entities": [! {! "type": "Company",! "relevance": "0.33",! "sentiment": {! "type": "positive",! "score": "0.639044"! },! "count": "1",! "text": "Google",! },! {! "type": "Company",! "relevance": "0.33",! "sentiment": {! "type": "positive",! "score": "0.639044"! },! "count": "1",! "text": "Gecko Design"! }! ]!}

Other Services

AI APIs

#deeplearningseries

Machine Translation Use it to: translate foreign-language content for your users, translate forum posts or chats, translate customer support emails

Question Answering

Machine Learning

Use it to: leverage questions like "was this company acquired?" to trigger processes in data pipelines or GUI displays, answer questions from your own data

Use it to: predict a prospect's likelihood to convert based on historical data, analyze customer churn

APP IDEA Customer Service - Enable customer support in any

language https://cloud.google.com/products/translate-api "

http://products.wolframalpha.com/api/ "

https://bigml.com/developers/ "

#deeplearningseries

EXAMPLE An AI-Driven App

Better understand customer feedback

The Challenge

As an application developer for a customer support center, I want to process thousands of emails so that I can:

• Determine how customers feel about our brand

• Understand how customers’ feedback evolves over time

• Discover interesting trends within sales and support data

• Develop meaningful reports for VP of Customer Support that show trends across time and product lines

#deeplearningseries

Build a customer support email analysis engine A Solution

What it will do: • Collect support email data • Process email content (uses AI) • Store data • Provide visual representation of

data for analysis

#deeplearningseries

Basic Workflow

Customer Support Email Analysis Engine

#deeplearningseries

#deeplearningseries

curl “https://www.googleapis.com/gmail/v1/users/alchemist@alchemyapi.com/messages/

147fe7dc5f9617e8?format=full&access_token={ACCESS_TOKEN}”

#deeplearningseries

curl “https://www.googleapis.com/gmail/v1/users/alchemist@alchemyapi.com/messages/

147fe7dc5f9617e8?format=full&access_token={ACCESS_TOKEN}”

curl -o sentiment.txt “http://access.alchemyapi.com/calls/text/

TextGetTextSentiment&apikey={API_KEY}&outputMode=xml&text=I just love using the services

provided by AlchemyAPI”

#deeplearningseries

curl “https://www.googleapis.com/gmail/v1/users/alchemist@alchemyapi.com/messages/

147fe7dc5f9617e8?format=full&access_token={ACCESS_TOKEN}”

INSERT INTO CustomerEmailSentimentTable (CustomerName, Domain, Text, Sentiment)

VALUES (‘WeLoveAlchemy’, ‘welovealchemy.com’, ‘I just love using the services provided by

AlchemyAPI’, ‘positive’)

curl -o sentiment.txt “http://access.alchemyapi.com/calls/text/

TextGetTextSentiment&apikey={API_KEY}&outputMode=xml&text=I just love using the services

provided by AlchemyAPI”

#deeplearningseries

Enjoy the analysis!

wget -O sentiment.txt “http://access.alchemyapi.com/calls/url/TextGetTextSentiment&apikey={API_KEY}&outputMode=xml&text=Google strengthens hardware expertise by acquiring Gecko Design”

INSERT INTO CustomerEmailSentimentTable (CustomerName, Domain, Text, Sentiment) VALUES (‘WeLoveAlchemy’, ‘welovealchemy.com’, ‘I just love using the services provided by AlchemyAPI’, ‘positive’)

curl “https://www.googleapis.com/gmail/v1/users/alchemist@alchemyapi.com/messages/147fe7dc5f9617e8?format=full&access_token={ACCESS_TOKEN}”

What We’ve Covered

#deeplearningseries

1

2

3

What’s Next? Get Started

•  Test deep learning with your own applications: Free API Key: http://www.alchemyapi.com/api/register.html

•  Get started with the guide: http://www.alchemyapi.com/developers/getting-started-guide/

•  SDKs available at: https://github.com/AlchemyAPI

•  Need help? Contact support@alchemyapi.com

Thursday, September 18 How Google Does It: Big and Fast

Semantic Analytics Pipelines

http://www.alchemyapi.com/how-google-does-semantic-analysis-webinar/

#deeplearningseries

What’s Next? APIs Discussed Today

•  AT&T Speech API - https://developer.att.com/apis/speech •  AlchemyVision - http://www.alchemyapi.com/products/alchemyvision/

•  AlchemyLanguage - http://www.alchemyapi.com/products/alchemylanguage/

•  Google Translate - https://developers.google.com/translate/ •  Wolfram Alpha - http://products.wolframalpha.com/api/

•  BigML - https://bigml.com/developers/ •  Visualization Services - https://developers.google.com/chart/,

http://matplotlib.org/contents.html

•  JSON Libraries - http://json.org/, https://docs.python.org/2/library/json.html

#deeplearningseries

#deeplearningseries

Contact us 1-877-253-0308

questions@alchemyapi.com www.alchemyapi.com

You will receive an email with a recording of this webinar, the slides and additional resources soon.

Thank you for attending!

#deeplearningseries

http://www.globalcraze.com/2014/01/29/install-google-translate-to-chrome/ http://www.headstudios.com.au/arnold-schwaznagger-talked-to-me/ http://www.oddcast.com/technologies/voicerecognition/ http://www.iphonefaq.org/archives/971714 http://cached.imagescaler.hbpl.co.uk/resize/scaleWidth/620/offlinehbpl.hbpl.co.uk/news/OKM/EEC22411-CD1A-3BD3-2BCC0331BEA06AE2.jpg

Images

#deeplearningseries