Stephen E. Robertson and M. M. Hancock-Beaulieu. On the evaluation of IR systems. Information Processing and Management, 28(4):457--466, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....found that, in theory, clustering the results of a query (rather than the whole collection) using Scatter Gather should be helpful for locating the relevant documents, as most of them are usually placed in the same cluster, in accordance with the cluster hypothesis. Kural, Robertson, and Jones [63] noted that the usefulness of the clustering of query results is dependent on how well the clusters are represented to the user, and they described the trade offs involved in producing a representation that is both compact and understandable. Hearst and Pedersen s experiment participants tended to ....

Y. Kural, S. Robertson, and S. Jones. Deciphering cluster representations. Information Processing and Management, 37(4):593--601, 2001.

.... a human intermediary, and the fact that there is usually no official login and logout to access and use the search engines, it is not very clear exactly what a unit session should be (at least from a system oriented definition perspective) Studying users in interactive systems is difficult [7], but studying users on the Web is even more difficult. Unless a special browser is used, the information available from the retrieval process about a user and his information need is much less for the Web search than a traditional library search. The queries are short [5] and with users ....

S. Robertson and C. Hancock-Beaulieu. On the evaluation of IR systems. Information Processing and Management, 28(4):457-476, 1992.

....UIIRS shares the difficulties of evaluating effectiveness of information retrieval systems (e.g. dealing with the evasive concept of relevance; coping with the core IR problems mentioned previously; variability of many factors, like systems, databases, intermediaries, etc. Tague Sutcliffe, 1992; Robertson and Hancock Beaulieu, 1992; Saracevic et al., 1988; Saracevic and Kantor, 1988a; 1988b] Moreover, it puts in the foreground users, their information problems, and their interaction with the UIIRS. The consequence is the need to deal not only with quantitative performance levels (such as precision and recall) but also with ....

....UIIRS. The latter objective stems from the need to evaluate different design options concerning not only the terminological support but also the more general conceptual support. 4.1. Experimental design The experiment was designed and executed together with a team of psychologists. According to [Robertson and Hancock Beaulieu, 1992], it can be classified as: A laboratory experiment (as opposed to an operational one) in which induced information needs were used and the bibliographic database (20,000 items regarding artificial intelligence derived from INSPEC) is smaller than real collections. A diagnostic experiment (as ....

S.E. Robertson and M. M. Hancock-Beaulieu, On the evaluation of IR systems. Information Processing and Management, 28(4), 1992, pp. 457--466

....linear case will follow by analogy. To show what is involved let me given an example of the estimation process using simple maximum likelihood estimates. The basis for our estimates is the following 2 by 2 table. j(i) i i j(i) x = 1 x = 0 [1] 2] 3] 4] 5] 6] x = 1 x = 0 [7] 8] [9] 98 Information retrieval Here I have adopted a labelling scheme for the cells in which [x] means the number of occurrences in the cell labelled x. Ignoring for the moment the nature of the set on which this table is based; our estimates might be as follows: i P (x = 1 x = 1) t i j(i) 1] ....

....I(x i ,x j ) s. The absolute values do not matter. Therefore if we use simple maximum likelihood estimates for the probabilities based on the data contained in the following table (using the same notation as on p. 125) j i i j x = 1 x = 0 [1] 2] 3] 4] 5] 6] x = 1 x = 0 [7] 8] [9] then I(x i ,x j ) will be strictly monotone with [1] log [1] 5] 7] 2] log [2] 6] 7] 3] log [3] 5] 8] 4] log [4] 6] 8] This is an extremely simple formulation of EMIM and easy to compute. Consider the case when it is P(x) we are trying to calculate. The MST is then ....

[Article contains additional citation context not shown here]

ROBERTSON, S.E., 'The probabilistic character of relevance', Information Processing and Management, 13, 247-251 (1977).

No context found.

Stephen E. Robertson and M. M. Hancock-Beaulieu. On the evaluation of IR systems. Information Processing and Management, 28(4):457--466, 1992.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC