No abstract

We introduce a framework of multiple viewpoint systems for describing and designing systems that use more than one representation or set of relevance judgments on the same collection. A viewpoint is any representational scheme on some collection of data objects together with a mechanism for accessing this content. A multiple viewpoint system allows a searcher to pose queries to one viewpoint and then change to another viewpoint while retaining a sense of context. Multiple viewpoint systems are well suited to alleviate vocabulary mismatches and to take advantage of the possibility of combining evidence. We discuss some of the issues that arise in designing and using such systems and illustrate the concepts with several examples. IntroductionSeveral different organizations of the same information often exist, and many data stores contain information that is related or complementary to other data store, but these different organizations are seldom integrated, and often are only used together in an ad hoc way so that a user must keep switching back and forth between different applications and interfaces in the course of a single search. Multiple viewpoints systems are an approach to describing systems that integrate more than one organizational approach across a body of information, gaining leverage in retrieval from their diversity and simplifying user interaction by uniting them under a common interface.This report describes a descriptive framework suitable for characterizing and discussing such systems. The framework provides a useful vocabulary for analyzing existing multiple viewpoint systems, as well as for addressing design decisions in the creation of new multiple viewpoint systems. Examples of several prototype systems developed using the multiple viewpoint framework will be introduced in Examples.In many cases, large amounts of data relevant to a problem or area of interest exist within a document 1 collection or other data store, but are inaccessible due to mismatches of vocabulary between searchers and the methods available for accessing this data. The inclusion of viewpoints using different vocabularies (or the same vocabulary used in different senses) in one system increases the likelihood that a searcher's vocabulary can be reconciled with the indexing vocabulary.A viewpoint is a representation of some set of relations that exist among a set of data elements, the domain set, along with a method for accessing this information, the access method. A valid member of the query language for the viewpoint, when presented to the access method, will elicit a result, a subset of the domain. This is called consulting the viewpoint. A system of multiple viewpoints consists of representations of several different sets of relational judgments about some collection of data. The relationship between two elements in a data collection may be strong according to one judgment and weak according to another-documents near each other alphabetically may be widely separated under the Dewey Decimal system-or simply noncomparabl...