英语论文网

er, no further directory with different orientations was listed so that it is difficult for users to identify the relevance between the query and the first results (Ryen W. White et al, 2006).
As to the introductory articles searching, many other options including “Year”, “Publication title” et al was available to refine your searching results at the beginning. For instance, 3912 results were found when “definition of nanomaterials” was input with detailed number distribution of different classification: year, publication title, topic and so on in refine filter. Therefore, the found result can be reduced to 14 when topic “nanotechnology” and year “2015” had been chosen make it more convenient for the users. When one article is chosen, recommended articles, citing articles together with the related book content were listed on the right side of the website to provide more options that the readers are interested in, which is considered as the reformulation process on query expansion.

4. Analysis 分析
As a global database, Elsevier is able to facilitate lots of users to search what they need effectively and accurately. However, there is still some space should be improved to make the service better. Take the known item searching as an example, the target results can be focused within a short time when the exact item was input. However, users can get nothing when a tiny error was typed. According to the research, more than 20% queries are misspelled due to various reasons (S. Cucerzan, 2004). As a result, the automatically correction of misspelled queries is urgent and necessary for the searching engines to develop and improve so that the information needs of the users can be met (Hargittai, 2006).
In general, the hierarchical faceted metadata mode that Elsevier applied on query refinement is well designed, including several metadata, such as author name, journal or book title, volume, issue and even page, which make it more precise for searching. At the same time, another level of refine filters involving “year”, “publication title” and “topic” et al could serve from another classification to broaden the users’ selection range, which can optimize the efficiency of searching (Ricardo & Berthier, 1999). What’s more, the query expansion function is well designed when one article or book is chosen under one query- three categories of expanding information will be listed as mentioned above. As a result, the retrieval performance of reformulate a query can be effectively improved (Olga Vechtomova, 2006). However, this kind of query expansion function does not show up before the articles or books list appears. Thus, the expanded queries are more related to the article that you choose from the list rather than the query that you input. This situation is acceptable if the article or book lists are highly consistent with the queries that users input. Nevertheless, what will happen if the relevance between the results and the queries is not so close? There may be a misleading to the users, which will result in waste of time.
Moreover, another potential drawback is also associated with the query reformulation process corresponded to the failed query search. When replacing the word “definition” by “introduction” from the query “definition of nanomaterials”, what we got form the results were only the articles or books with the key word “nanomaterials”, which indicated that it could be an