Aalbersberg I.J.: Incremental relevance feedback. In Proceedings of the Annual lnt. ACM S1G1R Conference on Research and Development in Information Retrieval, pp. 11 - 22, 1992  Home/Search   Document Not in Database   Summary   Related Articles   Check

....terminology or just one word more or less in his query then the traditional feedback environment doesn t recognize any similarities in these situations. Another idea to solve the terminology problem is to use query concepts. The system called Rule Based Information Retrieval by Computer (RUBIC) [1, 5, 18] uses production rules to capture user query concepts. In RUBIC, a set of related production rules is represented as an AND OR tree, called a rule base tree. RUBIC allows the definition of detailed queries starting at a conceptual level. The retrieval output is determined by fuzzy evaluation of ....

Aalbersberg I.J.: Incremental relevance feedback. In Proceedings of the Annual lnt. ACM S1G1R Conference on Research and Development in Information Retrieval, pp. 11 - 22, 1992

....based on relevance feedback from the Collection. Since the Collection feedback arrives continuously, the topic lter needs to be iteratively re ned. Incremental feedback techniques are fairly well known, but have typically been applied to situations where the underlying database remains constant [1]. Recent work, however, has focused on incremental feedback techniques for dynamic environments [2] The topic lter re nement algorithm employs the initial lter set by the Collection, Collection relevance judgements from past feedback cycles, and the new set of Collection judgments to generate ....

....was used as a crude upper bound on performance. The performance of a random algorithm, P = 11:0114, averaged over 200,000 iterations, was used as a lower bound. Results. Figure 3 illustrates the results for di erent values of the parameter in the range [0, 1) and the parameter in the range [0, 1] (see Sect. 3.3) The best performance of the harvesting strategy, achieved by = 0:4 and = 0:9, was P = 13:9814. Using the random algorithm as a lower bound, this performance was 0.71 times that of the o ine greedy algorithm. Further evaluation of this simple algorithm, particularly for ....

I. Aalbersberg. Incremental relevance feedback. In Proceedings of the Fifteenth Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 11-22, 1992.

....of generalized Euclidean distance, the query vector, and term weights. In practice, it is not feasible to require a user to provide feedback for a large number of retrieved images or documents. To solve this problem, incremental relevance feedback techniques were proposed by Aalbersberg [1] and Allan [3] These methods adjust term weights using only a few judged documents each time. Test results show that these methods can achieve good performance and is useful for handling long standing queries with drifting notions of relevance. Our method is similar to incremental relevance ....

....that is not already captured in the previous feedbacks. A feedback that does not provide new information is not useful for the algorithm and does not yield any improvement in the retrieval performance. This constraint is also true for any incremental relevance feedback method, such as those in [1, 3]. This retrieval experiment used the same images described in Section 4.1. Initially, the constructed space was initialized to a Euclidean space, and a support vector machine (SVM) was trained to map the Gabor features of the textures to the coordinates in the constructed space. For each ....

I. J. Aalbersberg. Incremental relevance feedback. In Proc. SIGIR '92, pages 11-22, 1992.

....and is utilized only when volunteered. Retrieval is adjusted through background sampling, anydata indexing, and dual space feedback. 1 Introduction The integration of query generation and user feedback continues to challenge information retrieval (IR) technologies. Relevance feedback [Aalbersberg 1992; Robertson and Walker 1997] the de facto standard for generating queries from user feedback, is useful to information science professionals [Spink and Saracevic 1997] but is frequently inappropriate for lay users. Many users are not able to adequately state their information needs in queries ....

Aalbersberg, I. J. Incremental Relevance Feedback. In Proceedings of the 15th Annual International SIGIR Conference, 1992.

....terminology or just one word more or less in his query then the traditional feedback environment doesn t recognize any similarities in these situations. Another idea to solve the terminology problem is to use query concepts. The system called Rule Based Information Retrieval by Computer (RUBIC) [1], 5] 12] uses production rules to capture user query concepts. In RUBIC, a set of related production rules is represented as an AND OR tree, called a rule base tree. RUBIC allows the definition of detailed queries starting at a conceptual level. The retrieval output is determined by fuzzy ....

Aalbersberg I.J.: Incremental relevance feedback. In Proceedings of the Annual Int. ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 11 - 22, 1992

....at Maryland and Brown [4] all contain a user profile management component. To date, however, these projects have not emphasized learning based acquisition and maintenance of profiles. There has been significant research on text based profile construction in information retrieval community (e.g. [5, 1, 7, 3]) especially in the framework of TREC [24] The main emphasis of TREC, however, has always been on the effectiveness of the participating systems, rather than on their efficiency. Most of the techniques used for these tasks require batch processing of previously judged documents, imposing ....

I. J. Aalbersberg. Incremental relevance feedback. In Proc. of the ACM SIGIR Conf., pages 11--22, Copenhagen, 1992.

....from the examined non relevant documents. This process is usually done in batches the documents in the collection are broken into training and testing sub collections, the queries are modified with the information from the training collection and evaluated on the test collection. Aalbersberg [1] explored the notion of incremental relevance feedback, where the documents are considered one at a time and the query is modified after each document. Allan [2] conducted an intensive study of the incremental approach showing results as good as if the feedback occurred in one pass. The most ....

I. J. Aalbersberg. Incremental relevance feedback. In Proceedings of ACM SIGIR, pages 11--22, 1992.

....in the relevance feedback, that is using the information provided by documents the user indicated as non relevant. In IR the use of negative data in relevance feedback has been received with contrasting views. Salton considered it positively [25] while other researchers considered it dangerous [1] or even harmful [11] We believe that it all depends on the particular retrieval model one is using. We intend to prove that our model make an effective use of negative data in relevance feedback and that the presence of negative data speeds up the learning of the parameters of a IF system. 3 A ....

I.J. Aalbersberg. Incremental relevance feedback. In Proceedings of ACM SIGIR, pages 11--22, Copenhagen, Danmark, jun 1992.

....that is using the information provided by documents the user indicated as non relevant. In IR the use of negative data in relevance feedback has been received with contrasting views. Salton considered it positively [Salton and McGill, 1983] while other researchers considered it dangerous [Aalbersberg, 1992]. Dunlop [Dunlop, 1997] observed that the vector space model [Salton, 1971] the term addition model [Harman, 1992] and the probabilistic model [Harper and van Rijsbergen, 1978, van Rijsbergen, 1979] behave differently for negative feedback. We intend to prove that the model makes an effective use ....

Aalbersberg, I. (1992). Incremental relevance feedback. In Proceedings of ACM--SIGIR 92 Conference, pages 11--22, Denmark.

....term. The same effect would be achieved if the activation was spread at run time from the term machine10 down to the term computer11. The client s feedback in retrieval When the client searches a Q A collection, the system uses another relevance feedback technique similar to the one proposed by Aalbersberg (1992). The system retrieves the Q A pairs one by one, as figure 3 shows. At each retrieval, the client can tell the system whether or not the retrieved Q A pair answers his or her question. The client can also request the retrieval of more Q A pairs without evaluating the quality of the pair that was ....

....has to catch up with the people, not vice versa. Most information retrieval systems use relevance feedback as the main method to support interaction with the client. But, while evaluations of relevance feedback have consistently shown improvements in performance (Ide Salton, 1971; Brauen, 1971; Aalbersberg, 1992), they have also consistently avoided assessments of their user friendliness. Typically, relevance feedback techniques require multiple retrieval interactions each of which asks the client to state which of the retrieved results are relevant to his or her question and which are not. Essentially, ....

[Article contains additional citation context not shown here]

Aalbersberg, I.J. (1992). "Incremental relevance feedback." Proceedings of the 15th Annual International SIGIR Conference, 1992 (pp. 11-21).

....interests that arise in an information filtering environment. The main difference of our work is the introduction of a parametric approach that adaptively changes the number of vectors used to represent profiles. Previously, Aalbersberg evaluated the effectiveness of incremental relevance feedback [Aal92], however, from the standpoint of an information retrieval environment. Lam et al. addressed the issue of shifts in user interests [LMMP96] using a two level approach which combines reinforcement and Bayesian learning. Unlike our work, which adopts a quite general definition of user profiles, ....

I. J. Aalbersberg. Incremental relevance feedback. In Proc. ACM SIGIR Conf., pages 11--22, 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