OR/16/047 Recommendations

Nkisi-Orji, I. 2016. Semantic information retrieval for geoscience resources: results and analysis of an online questionnaire of current web search experiences. British Geological Survey Internal Report, OR/16/047.

The following are recommendations on the implementation of semantic search features based on the findings of this survey.

• The ranking of relevant search results should focus on the top 20 results as most users may not assess results beyond that. This can provide a cut-off point when evaluating the performance of a search application.
• A feature which includes narrower or child terms of original search intent will be beneficial in a search application. Also, searches for equivalent terms or alternative spellings of original search intent will benefit users. These should be separate optional search features which users can turn on and off. Narrower/child terms are usually available in domain ontologies (thesauri, controlled vocabularies, etc.). Equivalent terms and alternative spellings can also be described in ontologies as alternative textual labels of concepts. The Geoscience thesaurus and Lexicon of Named Rock Units were identified as the most useful ontologies for this purpose.
• It will be beneficial to restrict search results to only return relevant documents, omitting irrelevant entries which users find are often returned. Irrelevant entries in search results can make it difficult to locate relevant documents. This is especially true when irrelevant results dominate the top ranks of search results.
• The resolution of ambiguous search terms will be beneficial to many users. It is preferable to present search application users with alternative meanings of ambiguous search terms from which they select intended meanings. The disambiguation of search terms should translate to the return of documents which express the intended context/meaning only.