In theory yes, and many search engines use this as a technique. Microsoft SharePoint search was (might still be) heavily reliant on tagging documents with metadata to support search. In practice however, and depending on the search application of the search engine, people don't always do it. Websites might contain domain specific keywords to support search engine optimisation, and there are tools to support doing this. In enterprise environments, however, you're relying on individuals both having the time to tag every document they create, and understanding the appropriate tags for said documents. One of the big challenges in enterprise search is that, for example, a finance department will used one set of domain specific terms compared to a design engineering department, both departments would also use the same information is different ways and for different purposes. If a design engineer creates a document, I'd argue that it's impractical to expect them to fully understand all the ways in which that document can be used, and all the things it could be called, across an entire organisation, particularly given that they are a design engineer and presumably paid to do design engineering, and not to create and disseminate content.
Depending on what your application, this paper might be of interest:
Mukherjee R, Mao J. Enterprise Search: Tough Stuff: Why is it that searching an intranet is so much harder than searching the Web?. Queue. 2004 Apr 1;2(2):36-46.
Besides my databases for notes and URLS, I also use a semantic network in the background. An obvious reason for a semantic network that it allows for finding database entries whose keywords are related to the keywords in the query.
A difficulty with tagging is that often you don't remember the keywords you attached to an article. Let's assume you have a note about a newly discovered 2D material, a honeycomb carbon lattice, and you have tagged this note with 'graphene'. When you search for the article again, you probably don't remember the word graphene. The semantic web can help you, in a first step, to recall the word graphene, which you can the use for database search.
Another reason for supporting search with a semantic network is that one can exploit the semantic distance between keywords for ranking results. Suppose you enter 'autonomous' and 'car' and 'vehicle'. Clearly, articles that contain both autonomous and car, autonomous and vehicle, should be ranked higher than articles that contain only car and vehicle - because the words car and vehicle have a small semantic distance. Semantic networks allow for emploing fuzzy logic in database search. Word pairs in a query that are semantically related should be OR-connected, word pairs that are semantically separated should be AND-connected.
Regards,
Joachim
- Badges
- Science topic
- Similar topics
- Engineering
- Materials Engineering
- Powders