HCI Lesson 7

Lesson 7 Overview

Study Guide

Message	Lesson 7 focuses on three emerging areas of Computer Science that have a large impact on Human-Computer Interactions -- Natural Language Processing, Deep Learning and Semantic Search. Computerized systems are increasingly engineered to act more like human beings to improve interactions with actual human beings. This lesson highlights some of the evolutionary advances that have taken place in recent years.
Reminders
Due Dates

Objectives

Understand advances in Natural Language Processing as presented in a recent NLP tutorial from the “Text by the Bay 2015” conference.

Become acquainted with Deep Learning by exploring some general HCI applications such as face recognition and a mobile application that translates images from one language to another.

Understand the principles, the challenges, and the underlying data structures of information retrieval
Explore the design considerations and usability of search engines

Understand the vision of the semantic web
Perceive how semantic search is implemented and analyze semantic improvements in information retrieval
Gain exposure to ontologies

Readings

Natural Language Processing
        Deep Learning: http://jackmyers.info/hci/lessons/07/deepLearning.html
        Natural Language and Word Vectors:
         http://jackmyers.info/hci/lessons/07/naturalLanguageAndWordVectors.html (video)
Deep Learning
Deep Learning: http://jackmyers.info/hci/lessons/07/deepLearning.html
Web 1.0 and Information Search
        Designing the User Interface, Chapter 13.
        From Information Search to Semantic Search (Part 1)
         http://jackmyers.info/hci/lessons/07/From_Info_Search_to_Semantic_Search.pptx

Web 3.0 and Semantic Search
        Evolution of the Worldwide Web from Web 1.0 to Web 4.0. by Aghaei, et.al.
         http://jackmyers.info/hci/readings/Aghaei_Web1toWeb4.pdf
        Building and Utilizing Ontologies for Knowledge Representation by Myers
         http://jackmyers.info/hci/readings/Myers_Ontologies.pdf

Instructions

Demo mode: which will use the "candle" / "butcher" example from the slides
Random mode: which will randomly generate term1 and term2 data structures (including random skip points and distances) based on a corpus of 1,000,000 words.

Your application should:

Produce the correct boolean results
Record the number of skips (Note: while the skip distances should be random from run to run, they should be constant for an individual run across both terms.)
Be very well commented

B. Implementing your own Ontology.
Download Protege and build an ontology of your own. (See the Fleetwood Mac
example in Building and Utilizing Ontologies for Knowledge Representation.

Make the ontology as complete as possible.
(For this, you might need to restrict the domain greatly.)

Did your ontology work as expected? What worked well? What did not work?

Another New Advance
Find a technical article or a blog on any recent software application, interaction device, or new technique/algorithm that affects HCI. Create a discussion post that includes the following:

A brief description of the content of the article/blog
At least one quote (please cite fully) from this source
What you think the pros/cons/challenges are for this technology from an HCI perspective
Your opinion on how much merit the solution holds

Also, you must reply to at least one other post from your classmates.
Utilize this discussion group. Please identify yourself when posting.

Search Engine Evaluations
Using Shneiderman’s “Five-Phase Framework” (pg. 521), evaluate five search engines against all of his bullet points. Two of the search engines must be Google and Bing, and at least one must be a specialized engine like Shakespeare Search. Be creative in your choice of the last three engines. Summarize your results.