We investigated the sources and effectiveness of search actions are still evolving, changing, and, at times, improvterms used during mediated on-line searching under ing. However, we still do not fully understand the many real-life (as opposed to laboratory) circumstances. A complex aspects of interactive processes, despite a numstratified model of information retrieval (IR) interaction ber of theoretical and experimental studies and scholarly served as a framework for the analysis. For the analysis, treatises (Ingwerson, 1992(Ingwerson, , 1996)). Furthermore, most of we used the on-line transaction logs, videotapes, and transcribed dialogue of the presearch and on-line interthe research and development in IR, concentrating on the action between 40 users and 4 professional intermediarimprovement of effectiveness in automatic representation ies. Each user provided one question and interacted with and searching, has treated IR systems and processes as one of the four intermediaries. Searching was done using static and not as dynamic or interactive (Saracevic, DIALOG. Five sources of search terms were identified: 1995). Such research, conducted now for 30 years, has (1) the users' written question statements, (2) terms derived from users' domain knowledge during the interacreached a certain maturity, as evidenced by the Text Retion, (3) terms extracted from retrieved items as reletrieval Conference (TREC) experiments (Harman, vance feedback, (4) database thesaurus, and (5) terms 1995). In contrast, research on the interactive aspects of derived by intermediaries during the interaction. Distri-IR has not reached maturity; it may be said to be emerging bution, retrieval effectiveness, transition sequences, and correlation of search terms from different sources were out of infancy. There is a clear need for two things: to investigated. Search terms from users' written question concentrate more research in IR on interactions, to more statements and term relevance feedback were the most resemble what is actually going on in practice, and to productive sources of terms contributing to the retrieval attempt to use what was found in interactive research for of items judged relevant by users. Implications of the findings are discussed.
* To whom all correspondence should be addressed. a mediated on-line IR interaction, to discern the sources of search terms and then to observe the effectiveness of