发明名称 Item selection in curation learning
摘要 A method of ranking candidate curation items includes receiving a query. The method also includes extracting items from multiple curations. The method also includes calculating, based on the query, a content similarity measurement for each of the extracted items. The method also includes extracting, from each of the extracted items, multiple curation-specific features. The method also includes calculating one or more curation-specific measurements for each of the extracted items based on the extracted curation-specific features, the one or more curation-specific measurements being different than the content similarity measurement. The method also includes ranking each of the extracted items based on both the content similarity measurement and the one or more curation-specific measurements to generate multiple curation item results.
申请公布号 US9454612(B2) 申请公布日期 2016.09.27
申请号 US201314013113 申请日期 2013.08.29
申请人 FUJITSU LIMITED 发明人 Wang Jun;Uchino Kanji
分类号 G06F17/30 主分类号 G06F17/30
代理机构 Maschoff Brennan 代理人 Maschoff Brennan
主权项 1. A method of ranking candidate curation items, the method comprising: receiving a query; extracting a plurality of items from a plurality of curations; calculating, based on the query, a content similarity measurement for each of the extracted plurality of items; extracting, from each of the extracted plurality of items, a plurality of curation-specific features; calculating one or more curation-specific measurements for each of the extracted plurality of items based on the extracted plurality of curation-specific features, the one or more curation-specific measurements being different than the content similarity measurement; ranking each of the extracted plurality of items based on both the content similarity measurement and the one or more curation-specific measurements to generate a plurality of curation item results; searching a plurality of web resources based on the query to generate a plurality of web results; searching a plurality of open education resources based on the query and learning-specific features to generate a plurality of open education resource results; unifying the plurality of web results, the plurality of open education resource results and the plurality of curation item results to generate a plurality of unified results including a plurality of items; receiving a selection of an item from the unified results; and adding the selected item to a curation; wherein the unifying the plurality of web results, the plurality of open education resource results and the plurality of curation item results to generate the plurality of unified results including the plurality of items comprises: extracting N top results from the plurality of open education resource results, and N top results from the plurality of curation item results, where N is a constant;adjusting an open education resource score of each of the N top results from the plurality of open education resource results based on an importance factor of the plurality of open education resource results;adjusting a curation item score of each of the N top results from the plurality of curation item results based on an importance factor of the plurality of curation item results; andmerging the N top results from the plurality of open education results with the N top results from the plurality of curation item results to generate merged results, each item in the merged results having a merged score, the merged scores of the items in the merged results being based on the corresponding adjusted open education resource scores and the corresponding adjusted curation item scores,wherein the plurality of unified results include N results from the merged results.
地址 Kawasaki JP