英语论文网

improper query. However, no related queries or relevant topic involved in the reformulation process had been presented in the search results. In my opinion, this is a distinct disadvantage for an IR system since the loss of reformulation process can hardly exhibit a user-centered feature, which may attribute to the decreasing attraction from the users. At the same time, the failed design on reformulation process can also limit the exploration of users in query expansion (Ruthven, 2008).

5. Recommendation 建议
As mentioned at the beginning of Analysis part, a distinct drawback of Elsevier system is the misspelling correction problem. No automatic correction or hint of incorrect query is shown when misspelling situation appears, which may lead to the loss of potential users after several failed attempts. As a consequence, I recommend that Elsevier should pay more attention on the new technology of automatically misspelled correction as well as the reminder function to a possible wrong query. In contrast, the corresponding function is well engaged in goggle scholar search engine, which owns large amount of users with fine experience (Duan, 2011).
Another recommendation is the navigation towards the hierarchical faceted metadata mode. For the IR system designers, how to support the hierarchical categories to facilitate the users has been an issue all the time. As described above, users can filter the results by refine the sub-categories. However, the number of results that users got is only the sum of individual number of the two sub-categories without other options, such as the results containing both sub-categories, leaving the users more tasks to seek their targets. Therefore, it is maybe an good idea that books and journals should be added with several key labels like the “key words” in the articles, which can provide convenience for the users to combine different topics (M. Hearst, 2006).
As to the reformulation process of Elsevier system, there are several drawbacks to be improved. Firstly, the query expansion function only exists associated to the results rather than the queries themselves. In other words, the relationship that IR system built is merely connected by the existing resources but not the potential results which users are searching for or interested in (Soo Young Rieh et al, 2006). Thus, this kind of experience won’t be widely accepted by lots users, especially the professional ones with specific purposes. As a consequence, the reformulation process towards the query expansion function built between queries and resources should be improved to meet the requirements of the users. In addition, another problem regarding to the reformulation process also exists as analyzed above. The system cannot present related queries or topics when unsuccessful queries are input at the beginning, which are able to bring much trouble for the users to proceed with their search just like the lack of misspelling correction. At this point, Google has to be mentioned as a positive model again. The combination of misspelling correction function with the related queries recommendations has been wonderful performed in google scholar search engine, which realize the perfect application of reformulation process, especially the query expansion. For the designers of Elsevier system, the misspelled correction technology together with eye-tracking technology (adopt three-line paragraph) may be hel