Author: Jorge Enrique Barreto

Design is an information intensive process. Designers base design decisions by gathering, synthesizing and analyzing information from various sources such as memos from colleagues or documented case histories. Translating design needs into information requests or queries aimed at finding the most useful information to help make design decisions represents a critical step in this process. Unfortunately, in the process of constructing information requests in the form of queries, designers must deal with several uncertainties, such as not being cognizant about the information available in the document database and/or not knowing how to express the needs in the query. This usually reduces the efficacy of the designers for finding the information most useful to their needs during the search process. Most information retrieval research has focused on retrieval techniques given a query rather than augmenting or re-writing queries to direct the search. Given that the information retrieval process is deciding whether or not to select a document based on the given decision variables, i.e., the terms in the query itself, there exists a need for a technique that applies a decision-theoretic framework for defining a search strategy to guide designers in formulating "intelligent" queries over full-text document databases. Through expected value of perfect information (EVPI) computation, the technique proposed in this thesis assists users in constructing queries which better represent their information needs because it directs the user towards terms which have the greatest potential of increasing the value of the decision, i.e., improving the amount of relevant information retrieved. The technique allows the user to add closely related keywords to words which have high EVPI and remove words which have low EVPI since there is no value in being more precise about the concept described by a particular low-value search term. The results found suggest that identifying those high-value attributes in the query improves decision-making, that is, the assurance of retrieving relevant documents because they closely match the information needs of the user.

Final version available in PostScript, and MS Word 7.0 for Windows

Authors: Andy Dong, J Enrique Barreto and Alice M Agogino

Information retrieval (IR) systems interact with users by returning a ranked list of relevant documents in response to a query. Through feedback mechanisms such as relevance feedback and automated keyword expansion, IR systems attempt to guide users in constructing search queries which better represent their information needs. These mechanisms, however, do not offer the user more insight into the content of the documents in the IR database nor do they provide direction as to which search terms might yield better search results in terms of relevance and certainty that the retrieved document contains the information the user intended to retrieve. This paper presents a methodology based on the decision-analytic concept of expected value of perfect information for controlling query augmentation in information retrieval. The system dynamically learns the content of the documents in the database to compute the utility (measured in terms of relevance) of retrieving certain documents in response to queries, where the words in the queries represent the random variables. By computing the expected value of perfect information for each query term, the system either suggests new search terms or suggests that the user terminate the search.

Submitted to the 1997 ACM SIGIR Conference. Draft versions available in PostScript, and MS Word 7.0 for Windows. Please do NOT re-distribute.