Cooper, W. S. (1968). Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19(1), 30--41.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....as possible (P R curve) 2) I want to find as many relevant documents as possible, but I m only willing to examine n documents (DCV curve) A better expression of the user s goals might be: 3) I want to find one relevant document or . 4) I want to find k relevant documents Cooper [14] introduces the concept of Expected Search Length (ESL) which measures the expected number of documents that must be examined to satisfy the user s need. If the documents are strongly ordered (no ties) the search length can be calculated directly. If there are ties, one must then estimate an ....

W.S. Cooper. Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30--41, 1968.

....target page u within the set of pages visited by crawler algorithm c for topic t. Note that if a crawler does not fetch a relevant page then the position of the page is taken by default to be MAX PAGES 1. This measure of the average crawl position is akin to the notion of expected search length [8] as adapted for crawling tasks [16] If the average position L is small, then on average the user will have to wait less time while non relevant pages are visited, in order to retrieve a relevant page. 3.3 Crawler Algorithms Based on a large set of crawlers evaluated in our previous work [17] as ....

W. Cooper. Expected search length: A single measure of retrieval e#ectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30--41, 1968.

....a P R curve. Based on the presentation of results at the recent TREC conference [1] it is fair to say that these two methods define the standard for evaluation in information retrieval. Other evaluation methods have been proposed, such as Swets OC curve [16] and Cooper s expected search length [5], but they have not gained widespread acceptance in the IR community. 2.2 Assumptions in Evaluation In a practical retrieval system, evaluation scores should correlate directly with search effectiveness. A successful search returns documents that satisfy the information need of the user. Before ....

W.S. Cooper. Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30-- 41, 1968.

....of the search process. The average search length (ASL) is the expected position of a relevant document in the ordered list of all documents (Losee, 1998) This is related to the expected search length (ESL) the number of non relevant documents retrieved up to a certain point in the search (Cooper, 1968), and the ranked half life indicator, which represents the median position for the relevance frequency or relevance values for documents preceding a cutoff point (Borlund Ingwersen, 1998) The ASL is measured in units of documents ; knowing that the average position of a relevant document is 23 ....

Cooper, W. S. (1968). Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19(1), 30--41.

....2: Comparison of learned systems and individual search queries. 6. Related Work Problems that involve ordering and ranking have been investigated in various elds such as decision theory, the social sciences, information retrieval and mathematical economics (Black, 1958; Kemeny Snell, 1962; Cooper, 1968; Fishburn, 1970; Roberts, 1979; Salton McGill, 1983; French, 1989; Yao, 1995) Among the wealth of literature on the subject, the closest to ours appears to be the work of Kemeny and Snell (1962) which was extended by Yao (1995) and used by Balabanov c and Shoham (1997) in their FAB ....

Cooper, W. (1968). Expected search length: A single measure of retrieval eectiveness based on the weak ordering action of retrieval systems. American Documentation, 19, 30-41.

....700 nodes, and the associated graph of approximately 11,000 articles. 4. 1 Macro analysis In the first experiment we measure one version of search length, a criterion combining recall and precision, defined as the number of non relevant documents visited until the first relevant document is found [4]. This measure provides us with a macro analysis because the collective behavior of the agent population is observed as a whole, and averaged across a large number of queries. The search length of ARACHNID is compared with that of a simple non adaptive algorithm, namely BFS ....

W. Cooper. Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30--41, 1968.

....respectively. However, in our scenario we are not concerned with locating every relevant item. Rather, we are concerned with how much non relevant data a user must go through before finding some number of relevant items. This value can be measured by what is called expected search length (ESL)[2]. Expected search length measures the number of false hits a user or system would have to read through or process before finding a specified number of relevant items. 5 Experiment Results We ran SPIRE using three problem cases and collected the top documents for each. Removing duplicates and ....

William S. Cooper. Expected Search Length: A Single Measure of Retrieval Effectiveness Based on the Weak Ordering Action of Retrieval Systems. American Documentation, 19:30--41, 1968.

....in the rankings. Further, we want a single value so that comparison is easily done. Therefore, our primary concern is with how much non relevant data a user must go through before finding some number of relevant items. This value can be measured by what is called the expected search length (esl)[10]. ESL measures the number of false hits a user (or system) would have to read through (or process) before finding a specified number of relevant items, q. It measures the amount of wasted effort. Using this measure allows that there may be multiple items with the same probability of retrieval (in ....

Cooper, William S. Expected Search Length: A Single Measure of Retrieval Effectiveness Based on the Weak Ordering Action of Retrieval Systems. American Documentation 19 (1968), 30--41.

....by examining the passages retrieved at certain cutoff depths, we will lose information about the ordering of the relevant and non relevant passages. We are concerned with how much effort will be wasted by users as they examine retrieved passages. This can be measured by expected search length (esl)[Coo68], which measures the number of non relevant items encountered before finding a specified number of relevant ones. In this work we use esl 1 , esl 3 , and esl 5 , which are esl values when 1, 3, or 5 passages are specified. 5 Experiment results We ran SPIRE using three problem cases and collected ....

William S. Cooper. Expected Search Length: A Single Measure of Retrieval Effectiveness Based on the Weak Ordering Action of Retrieval Systems. American Documentation, 19:30--41, 1968.

....as a control parameter [52] Another metric lends itself better to assess the performance of on line retrieval systems. Search length is normally defined as the number of irrelevant documents that appear, in some ordered list of retrieved documents, in front of some fraction of the relevant set [7]. We can easily extend this method by imagining that only visited documents appear in the list of retrieved documents, and that their ordering is given by visit time rather than rank length then refers to waiting time. For an on line retrieval system this means that we only need to wait until ....

WS Cooper. Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30--41, 1968.

....above, where we simply counted the number of good passages in the top n. We should also look at 6. where we see if there are queries in which we have not satisfied the requirement to find any of the acceptable passages. One measure that incorporates these features is Expected Search Length (ESL) Coo68] It measures the amount of effort wasted when trying to find a particular number of relevant items while searching through a ranked list. From ESL we can compute the average ESL to compare values across queries when the collection and retrieval engine remain stable. Further, we can derive the ....

William S. Cooper. Expected Search Length: A Single Measure of Retrieval Effectiveness Based on the Weak Ordering Action of Retrieval Systems. American Documentation, 19:30--41, 1968.

....respectively. However, in our scenario we are not concerned with locating every relevant item. Rather, we are concerned with how much non relevant data a user must go through before finding some number of relevant items. This value can be measured by what is called expected search length (esl)[Coo68]. Expected search length measures the number of false hits a user or system would have to read through or process before finding a specified number of relevant items. Simply, it is the amount of wasted effort. 5 Experimental results We used SPIRE to retrieve relevant documents for 3 different ....

William S. Cooper. Expected Search Length: A Single Measure of Retrieval Effectiveness Based on the Weak Ordering Action of Retrieval Systems. American Documentation, 19:30--41, 1968.

....I vote the young lady tells us a story. I m afraid I don t know one, said Alice, rather alarmed at the proposal. Problems that involve ordering and ranking have been investigated in various fields such as decision theory, the social sciences, information retrieval and mathematical economics [5, 21, 8, 13, 26, 27, 14, 35]. Among the wealth of literature on the subject, the closest to ours appears to be the work of Kemeny and Snell [21] which was extended by Yao [35] and used by Balabanov ic and Shoham [1] in their FAB collaborative filtering system. These works use a similar notion of ranking functions and ....

W.S. Cooper. Expected search length: A single measure of retrieval effectiveness based on the weak ordering action of retrieval systems. American Documentation, 19:30--41, 1968.

No context found.

Cooper, W. S. (1968). Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19(1), 30--41.

No context found.

William S. Cooper. Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19(1):30--41, 1968.

No context found.

W. S. Cooper. Expected search length: A single measure of retrieval e#ectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30--41, 1968.

No context found.

Cooper, W.S. (1968). Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19, 30 -- 41.

No context found.

W. Cooper. Expected search length: A single measure of retrieval effectiveness based on the weak ordering action of retrieval systems. American Documentation, 19(1):30--41, 1968.

No context found.

Cooper, W. (1968). Expected search length: A single measure of retrieval effectiveness based on the weak ordering action of retrieval systems. American Documentation, 19(1), 30--41.

No context found.

W. Cooper. Expected search length: A single measure of retrieval effectiveness based on the weak ordering action of retrieval systems. American Documentation, 19(1):30--41, 1968.

No context found.

W. S. Cooper. Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30--41, 1968.

No context found.

W. S. Cooper. Expected search length: A single measure of retrieval e#ectiveness based on weak ordering action of retrieval systems. Journal of the American Society for Information Science, 19:30--41, 1968.

No context found.

W. Cooper. Expected search length: A single measure of retrieval effectiveness based on the weak ordering action of retrieval systems. American Documentation, 19(1):30--41, 1968.

No context found.

W.S. Cooper. Expected search length: A single measure of retrieval eectiveness based on the weak ordering action of retrieval systems. American Documentation, 19:30-41, January 1968.

No context found.

COOPER, W.S., 'Expected search length: A single measure of retrieval effectiveness based on weak ordering action of retrieval systems', Journal of the American Society for Information Science, 19, 30-41 (1968).

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