D13-1045 log of 15 million queries from a commercial search engine . They found that 90 % of queries
D09-1055 feature set from a state-of-the-art commercial search engine . This set includes NavClick
D10-1110 Timestamp queries . We scraped a commercial search engine using the 800 queries . We extracted
D15-1054 features that are typically used in commercial search engines . The first feature type is based
D14-1117 balancing were used . We believe commercial search engines can cut this down to less than
D14-1002 each entity name as a query to a commercial search engine , and retain up to the top-100
D15-1054 the click prediction system of a commercial search engine . Our evaluation results clearly
D09-1068 not be ignored in the case of commercial search engines . For this reason , we restrict
D13-1102 from the queries submitted to a commercial search engine during a week in mid-2012 . Every
D09-1083 query/suggestion was first issued to a commercial search engine . The result page with up to
D13-1045 used the ranking returned by a commercial search engine as our one of the Baselines .
D13-1045 100 0.54 CG ND results from a commercial search engine . The Measurement documents were
D13-1045 the original search results of a commercial search engine . This means that using high-quality
D10-1121 use the " near " operator from a commercial search engine on a given word and a seed word
D15-1054 reduce AucLoss by up to 4.1 % . For commercial search engines which have a very strong baseline
D09-1083 implement a paid-search ad filter for commercial search engines . In this scenario , textual
C04-1147 on the statistics provided by commercial search engines ( Zhu and Rosenfeld , 2001 ;
D09-1068 top ranked documents returned by commercial search engines for these queries ; this set
D15-1054 sponsored search data set from a commercial search engine , and demonstrate that our proposed
D09-1083 way . From the search logs of a commercial search engine , a random sample of 363 thousand
hide detail