Supporting Effective Access through User- and Topic-Based Language Models

1

• Supporting Effective Access through User- and
  Topic-Based Language Models

2
Who we are


3
Specifics

• People
• Nick Belkin, Principle Investigator, Rutgers
• Bruce Croft James Allan, Principle
  Investigators, U Mass
• Diane Kelly, Graduate Assistant, Rutgers
• Jeong Hyun (Annie) Kim, Yang-woo Kim, Hyuk-Jin
  Lee, Anne Washington, Rutgers
• Funding
• NSF 99-11942, Rutgers awarded 243,771 for a
  3-year period

4
The Problem

• Information retrieval and filtering tools make
  mistakes that are obvious and aggravating to
  users.
• Queries are often totally misinterpreted by
  search tools and, even when relevant documents
  are found, they are usually mixed with many
  others that are totally unrelated.
• WHY? Lack of adequate models of the user and of
  the domain.

5
General Solution

• Language Models
• Assume that associated with every document or
  group of documents there are one or more
  probability distributions that model how the text
  in the document can be generated.
• Can be viewed as models of the important topics
  that are covered by a document or a group of
  documents.

6

• Have the potential for representing topics of
  interest to a user or a group of users.
• Assume that different types of information
  problems are expressed in language in different
  ways. These different ways could be characterized
  by language models understood as models of the
  user.

7
Types of Language Models

• Topic-based
• Identify characteristics of the language that is
  used in a particular domain or topic.
• User-based
• Identify characteristics of the language that a
  user or a group of users associate with a
  particular domain or topic.

8
Overview of the Retrieval Process
9
An Example

• Consider the following information needs that
  indicate a users interest in Java
• I want to find the best and least expensive way
  to have a 2 week holiday in Java.
• I need to know where Java is.
• I need to write a brief survey of the current
  political situation in Java.

10

• I need to learn the programming language Java.
• I want to buy some dark roast Java coffee.
• Consider the following premise Documents that
  are relevant to each of these kinds of queries
  will have different language characteristics
  associated with them.

11
Our Goal

• To identify the language characteristics of each
  of these types of needs (user)
• To identify the language characteristics of the
  groups of documents that satisfy each of these
  types of needs (topic)
• To match these needs and document clusters
  accordingly