Nick Craswell, David Hawking, and Stephen Robertson. Effective Site Finding Using Link Anchor Information. SIGIR'01, pages 250-257, 2001.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....ignored for the Ad Hoc task, since they were not effective. Our experiments for the EP task at TREC 2001 have shown that link structure, URLs and anchor texts are useful sources of information for locating entry pages[38] Experiments of other groups confirmed the effectiveness of these features [6, 14, 7, 27, 32, 39]. In this paper we show how knowledge about the relationship between the non content features of a webpage an its likelihood of being an EP can easily be incorporated in a retrieval model based on statistical language models. In Section 2 we discuss related work on tuning retrieval models to a ....

....0.4920 P (Q D)P inlink (D) 0. 5963 Table 6: Results for different priors(content anchor) In section 4 we saw that both the number of inlinks and the URL can give important information about whether a page is an entry The experiments in this paper are based on the CSIRO anchor collection [6, 7], whereas the experiments reported in [38] are based on an anchor collection we built ourselves page or not. The results in Table 4 confirm this. A combination of these two sources of information might also be useful. In section 3.2 we discussed different approaches for combining priors. Table 8 ....

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In Croft et al. [9], pages 250--7.

.... links are particularly useful [8] There have been a couple of victories for link methods: Bailey, Craswell Hawking showed that they could be useful for site finding tasks [2] and Craswell, Hawking Robertson also obtained good results for site finding tasks, using link anchor text information [4] rather than the links themselves. Use of link anchor text may be one of the reasons that search engines using link analysis methods, such as Google, seem to perform so well even though there is no documented performance benefit from combining term and link information. Ideally we would have used ....

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In Proc. 24th SIGIR, pages 250-257, 2001.

....an information retrieval system, although the IR package is a satisfactory tool in retrieving news articles. Thus, we need to develop some revised methods of retrieving unarranged documents. In this study, we did not use link information or an interactive mode of the IR package. Previous research [4] has shown the effectiveness of using link information. We would like to investigate the approach of using link information. Results of average precision for the LARGE collection are worst than for the SMALL collection. It can be considered to affect elimination of documents badly. We should cope ....

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In SIGIR 2001.

....task was TREC 2001 s home page finding task [9] For entry page finding, non content features such as URLs and links provided valuable information [11] We did not see a straightforward way to use non content features for this year s task. An alternative is to use the anchortexts in the collection [5]. For the named page finding task, we experimented with plain text runs, anchor text runs, and their combinations. Table 7: Overview of the named page finding runs. 1. UAmsT02WnTl Text only Lnu.ltc 2. UAmsT02WnTm Text only Lnm.ltc 3. UAmsT02WnA Anchors only Lnu.ltc 4. UAmsT02WnTlA Combined ....

....667,737 documents, which is 54 of the collection. Third Anchors Index We use the same procedure as for the second anchors index, but now retain all links as they appear in the collection. Thus, if the same anchor text occurs thousands of times, we include it thousands of times (similar to [5]) The resulting index is based on 2,766,946 anchor texts covering 667,737 documents, which is 54 of the collection. Table 11: Anchors only run results. Run Index MRR Prec. at 10 UAmsT02WnA Anchors 1. 0.1391 UAmsT02WnA Anchors 2. 0.3279 UAmsT02WnA Anchors 3. 0.3098 UAmsT02WtA Anchors 1. ....

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In Kraft et al.

.... inspired by others who have highlighted how mediation and referral can lead to emergent communities in peer networks [37] The use of referential text to identify semantic similarity between pages is also not new [12, 2, 17] In addition, the idea has also been used for example to find Web sites [16], categorize pages [4] crawl pages [28] and rank crawled pages [23] Despite the active use of referential text for variety of information retrieval tasks, no one has yet demonstrated the effectiveness of this technique for generalpurpose search. There is a large and growing body of work on ....

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In Proc. 24th Annual Intl. ACM SIGIR Conf. on Research and Development in Information Retrieval, 2001.

....ignored for the Ad Hoc task, since they were not effective. Our experiments for the EP task at TREC 2001 have shown that link structure, URLs and anchor texts are useful sources of information for locating entry pages[38] Experiments of other groups confirmed the effectiveness of these features [6, 14, 7, 27, 32, 39]. In this paper we show how knowledge about the relationship between the non content features of a webpage an its likelihood of being an EP can easily be incorporated in a retrieval model based on statistical language models. In Section 2 we discuss related work on tuning retrieval models to a ....

.... 0.7748 4 K 0. 5963 Table 6: Results for different priors(content anchor) In section 4 we saw that both the number of inlinks and the URL can give important information about whether a page is an entry The experiments in this paper are based on the CSIRO anchor collection [6, 7], whereas the experiments reported in [38] are based on an anchor collection we built ourselves page or not. The results in Table 4 confirm this. A combination of these two sources of information might also be useful. In section 3.2 we discussed different approaches for combining priors. Table 8 ....

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In Croft et al. [9], pages 250--7.

....Experiment using WT10g We conducted a small experiment to determine whether there is sufficient link information within WT10g to enhance home page finding effectiveness 4 . Home page finding is an important Web search problem in which link information has proven useful in previous experiments (Craswell et al. 2001) and in commercial search engines. We randomly chose 100 site home pages from WT10g, and manually generated a query (a name for the site) for each. For example, the query Vidiot was generated for the home page http: www.cdsnet.net vidiot . The query processing task demands that the required ....

Craswell, N., Hawking, D., and Robertson, S. (2001). Effective site finding using link anchor information. In submission.

No context found.

Nick Craswell, David Hawking, and Stephen Robertson. Effective Site Finding Using Link Anchor Information. SIGIR'01, pages 250-257, 2001.

No context found.

Nick Craswell, David Hawking, and Stephen E. Robertson. Effective site finding using link anchor information. In Proc. of the 24th annual international ACM SIGIR conference on research and development in information retrieval, pages 250--257, New Orleans, Louisiana, USA, September 2001. Association for Computing Machinery.

No context found.

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In Research and Development in Information Retrieval, pages 250--257, 2001.

No context found.

N. Craswell, D. Hawking and S. Robertson, Effective site finding using link anchor information, ACM SIGIR, New Orleans, 2001.

No context found.

Craswell, N., Hawking, D., Robertson, S.E. Effective site finding using link anchor information. ACM-SIGIR'2001, 250-257.

No context found.

Craswell N., Hawking D., and Robertson S. Effective site finding using link anchor information. In Proc. of SIGIR'01, 250-257, September 2001.

No context found.

N. Craswell, D. Hawking, and S. Robertson. Effective site finding using link anchor information. In Proceedings of the 24 International ACM SIGIR Conference on Research and

No context found.

Craswell, N., etc., Effective site finding using link anchor information. In S/G/R-01.

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