Raghavan, V., Bollmann, P., and Jung, G. (1989). A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205-229.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of precision and recall values from single queries by averaging precision values at constant recall values to obtain the well known monotonically decreasing average PR curves without paying attention to the generality or window size values associated with those measurements. In the critical review [10] the authors state with respect to averaging precision and recall values within the same database that precision values should be averaged by using constant scope or window values rather than using constant recall values. The fact that results for equal cluster related window sizes lay along a ....

V. V. Raghavan, G. S. Wang, and P. Bollmann, A critical investigation of recall and precision as measures of retrieval system performance, ACM Trans. Inf. Syst., Vol 7, No 3, 2052.

.... t is classified into this class. Between high recall and high precision exists a trade off. All methods examined in this paper make category assignments by thresholding a confidence value . By adjusting this threshold we can achieve different levels of recall and precision. The PRR method [Raghavan et al. 1989] is used for interpolation. Since precision and recall are defined only for binary classification tasks, the results of multiple binary tasks need to be averaged to get to a single performance value for multiple class problems. This will be done using microaveraging [Yang, 1997] In our setting ....

Raghavan, V., Bollmann, P., and Jung, G. (1989). A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205-229.

....the assessor, for example, has not simply checked whether the query terms occur in each document) they should not in the general case introduce bias into measurement of the relative performance of systems. Similarly, given assessments there are many techniques for assigning a score to a system [8, 9, 10, 12], which can be based on theoretical considerations or pragmatic assumptions concerning the purposes of a system. Other aspects of the experimental methodology are, Permission to make digital hard copy of all or part of this work for personal or classroom use is granted without fee provided that ....

V.V. Raghavan, G.S. Jung, and P. Bollman. A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205--229, 1989.

....includes automatic spelling correction. Both the test suites of name matching algorithms made in [Pfeifer et al. 1995] and [Zobel Dart 1996] measure these criteria. The specific form of Precision Recall graphs used in my tests, is employed in the former study and recommended in for example [Raghavan et al. 1989]. The number of correct matches completely ignored by a matching algorithm gives a complementary view of the retrieval quality and a similar method is used in [Zobel Dart 1996] Good speed performance finally, is of utter importance for a commercial look up application. 33 Precision C B ....

....long as the values are calculated for integer values of matches. That is when you have found a new correct match you can find the corresponding precision and recall values. A problem arises during the interpolation process for weakly ranked algorithms. The Probability of relevance, PRR, defined in [Raghavan et al. 1989] gives a valid interpretation of both precision and recall also for noninteger numbers of retrieved documents. This method has been used with the n gram and edit distance methods, while single value precision and recall was used for the unranked Soundex based algorithms. Using the following ....

. Raghavan, Vijay V., Jung, Gwang S., and Bollmann, Peter. A Critical Investigation of Recall and Precision as Measures of Retrieval Systems Performance, ACM Transactions on Information Systems, Vol 7 (3), July 1989, pp 205-229

.... a single measure using the various parameters for evaluating a retrieval system has proven 19 Relevant Documents Relevant Documents Answer Set R A in Answer Set Ra Collection Figure 11: Recall and Precision for a sample answer set (adapted from [69] to be extremely difficult [41]. We would like to evaluate the system in a comparative way to measure how much do certain changes, such as varying the normalized number of colors in an image or the length of the binary signature bit string lead to an improvement in performance. The effectiveness of an information retrieval ....

....ordering among all the documents. From a user s perspective, a retrieval system should retrieve as many relevant documents as it can, with minimum number of non relevant ones. Roughly speaking, the former corresponds to the concept of recall, while the latter pertains to the notion of precision [41]. The Recall of an information retrieval system refers to the percentage of the total relevant documents retrieved [6, 67] and is mathematically defined as the total number of retrieved relevant documents from all the relevant documents. Figure 11 shows a graphical representation for a sample ....

[Article contains additional citation context not shown here]

V. Raghavan, P. Bollmann, and G. S. Jung. A critical investigation of recall and precision as measures of retrieval system performance. In ACM Transactions on Information Systems, pages 205--229, 1989.

....on the two well know statistics recall and precision widely used in information retrieval. Precision is the probability that a document predicted to be in class truly belongs to this class. Recall is the probability that a document belonging to class is classified into this class (see [Raghavan et al. 1989]) Both can be estimated from the contingency table. Between high recall and high precision exists a tradeoff. The P R breakeven point is defined as that value for which precision and recall are equal. The transductive SVM uses the breakeven point for which the number of false positives equals the ....

Raghavan, V., Bollmann, P., and Jung, G. (1989). A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205--229.

....cf. section 2.1 above) uses 10 For a motivation for this particular evaluation measure see [KP96b] p.44. 11 The results for c4fow and vn are slightly different from those reported in [KP96b] This is a result of using a different interpolation method (probability of relevance, cf. [RJ89]) for calculating recall precision values. version avp change ap5 15 change vMa1 0.34972 (0.21535) 11.86 0.44290 (0.28363) 14.25 vMa2 0.33997 (0.22486) 8.74 0.44815 (0.30933) 15.61 vMa3 0.34374 (0.22668) 9.95 0.45093 (0.31084) 16.32 vMa4 0.32962 (0.21242) 5.43 0.42654 ....

Vijay V. Raghavan and Gwang S. Jung. A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205--229, 1989.

.... the quality of a view rewriting, we need to estimate sizes of overlapping view extents in order to determine how much information is retained by a new query and how much meaningless new data is introduced [LKNR98] This parallels the concept of precision and recall from information retrieval [RJB89] though we now apply these concepts 14 Lee, Koeller, Nica, and Rundensteiner to the relational database context by addressing for example extent subset estimation. 7 Conclusion The concept of relaxed query semantics creates a new problem of non equivalent query rewritings that have to be ....

V. V. Raghavan, G. S. Jung, and P. Bollmann. A Critical Investigation of Recall and Precision as Measures of Retrieval System Performance. ACM Transactions on Office Information Systems, pages 205--229, July 1989.

.... the quality of a view rewriting, we need to estimate sizes of overlapping view extents in order to determine how much information is retained by a new query and how much meaningless new data is introduced [LKNR98] This parallels the concept of precision and recall from information retrieval [RJB89] though we now apply these concepts to the relational database context by addressing for example extent subset estimation. Much research has been done on query reformulation using materialized views. Levy et al. LRU96, LMS95, SDJL96] consider the problem of replacing an original query with a ....

V. V. Raghavan, G. S. Jung, and P. Bollmann. A Critical Investigation of Recall and Precision as Measures of Retrieval System Performance. ACM Transactions on Office Information Systems, pages 205--229, July 1989.

....Chapters in both Salton and McGill [75] and van Rijsbergen [70] present good overviews of evaluation in IR. Special issues in both Information Processing and Management [40] and Journal of American Society of Information Science [85] have focused on more recent evaluation trends. Raghavan et al. [69] examine recall and precision in particular and identify situations leading to ambiguous results. Tague Sutcliffe and Blustein [84] look specifically at evaluation in the Text Retrieval Conferences [43] Harter [46] and Ellis [24] are the latest in a long line of researchers who have questioned ....

V. V. Raghavan, G. S. Jung, and P. Bollmann. A Critical Investigation of Recall and Precision as Measures of Retrieval System Performance. ACM Transactions on Information Systems, 7(3):205--229, 1989. Bibliography143

....a view rewriting, we need to estimate sizes of overlapping view extents in order to determine how much information is retained by a new query and how much meaningless new data is introduced. This in some way parallels the concept of precision and recall used in the field of information retrieval [RJB89] although it is set in an entirely different context. Information retrieval generally does not deal with selecting subsets of tuples from a typed relation nor with combining such relation fragments via joins into larger result tuples. Rather, the work on precision and recall establishes measures ....

V. V. Raghavan, G. S. Jung, and P. Bollmann. A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Office Information Systems, pages 205--229, July 1989.

No context found.

Raghavan, V., Bollmann, P., and Jung, G. (1989). A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205-229.

No context found.

V. Raghavan, P. Bollmann, and G. Jung. A critical investigation of recall and precision as measures of retrieval system performance. In ACM Transactions on Information Systems, 7(3):205-229. 53

No context found.

V. Raghavan, P. Bollmann, and G. Jung, `A critical investigation of recall and precision as measures of retrieval system performance', ACM Transactions on Information Systems, 7(3), 205--229, (1989).

No context found.

V. V. Raghavan, P. Bollmann, and G. S. Jung. A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205--229, 1989.

No context found.

Raghavan, V.V., Bollmann, P., & Jung, G.S. (1989). A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3), 205--229.

No context found.

Springer. V. V. Raghavan, P. Bollmann, and G. S. Jung. A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205--229, 1989.

No context found.

Vijay Raghavan, Peter Bollmann, and Gwang S. Jung. A critical investigation of recall and precision as measures of retrieval system performance. ACM Trans. Inf. Syst. 7, 3 (July 1989), Pages 205-229.

No context found.

V. V. Raghavan, G. S. Wang, and P. Bollmann, A critical investigation of recall and precision as measures of retrieval system performance, ACM Trans. Inf. Syst., Vol 7, No 3, 205-229, 1989.

No context found.

Raghavan, V., Jung, G., and Bollman, P., "A Critical Investigation of Recall and Precision as Measures", ACM Transactions on Information Systems 7(3), page 205-229.

No context found.

Raghavan, V., Jung, G., and Bollman, P., "A Critical Investigation of Recall and Precision as Measures", ACM Transactions on Information Systems 7(3), page 205-229, 1989.

No context found.

V. V. Raghavan, P. Bollmann, and G. S. Jung. A critical investigation of recall and precision as measures of retrieval system performance. ACM Transactions on Information Systems, 7(3):205--229, 1989.

No context found.

Raghavan, V., Jung, G., and Bollman, P., "A Critical Investigation of Recall and Precision as Measures", ACM Transactions on Information Systems 7(3), page 205-229, 1989.

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