There are several reasons that educators and/or learners cannot find the content the learning content they would like find on the web. Some of those reasons mean that even this suggestion will not solve the problem but it may provide some interim steps to improve discovery until some of the underlying causes are remedied. For the purpose of this suggestion I will temporarily ignore the issues of content that is not ...more »
There are several reasons that educators and/or learners cannot find the content the learning content they would like find on the web. Some of those reasons mean that even this suggestion will not solve the problem but it may provide some interim steps to improve discovery until some of the underlying causes are remedied.
For the purpose of this suggestion I will temporarily ignore the issues of content that is not exposed to direct indexing and searching (aka dark web) and the lack of suitable metadata, and instead, just suggest an approach and play out some likely scenarios.
We can't find the content we need because we cannot construct the right queries to find it. Most people are not particularly adept at creating syntactically robust queries that use operators to refine the query, and therefore, return better results. Providing a search form for query creation that is better suited to learning content could help many people to find the content for which they are searching. Simple keyword queries are a pretty inaccurate "shotgun" approach to finding content. Even when results are returned, there is seldom sufficient information in the sample text that is returned and the searcher must visit the actual resources found to get some idea of their usefulness. Typing in questions provides greater accuracy because there are more words to associate, but even so, the results are not ideal.
If searchers could fill in a form that contained relevant options to educational searches then more accurate results would be returned. Aside from just the subject area or course name itself, selections for the level of education (primary, secondary, vocational, corporate training, etc), along with a reference to the year level or prerequisite knowledge, free/commercial/any, perhaps selections such as level of modularity (complete course vs module of a course) and perhaps even a preferred institution/initiative/vendor, any accessibility requirements, license restrictions etc etc. Then when the results are returned, provide a way to filter on the properties that were included in the search. Even if the search did not include all the fields, the results list could still be categorised along the lines of the fields suggested. These two simple steps could quite easily be implemented by search engines with some quite reasonable benefits.
Including these sorts of criteria would no doubt influence indexing in the future. Over time, accuracy would improve, however, until the metadata required to provide reliable results is available we would still have to rely on indexing and algorithms to interpret the nature of the content (not true semantic analysis just what is currently done).
A better implementation would include changes in the form based on some of the values that get filled in or even via user preferences. There is a lot more that could be said but this gives the gist of the suggestion and how it would help.
It has already been mentioned that custom search engines already exist but that these fragment the act of searching. Potentially this suggestion could have the same effect. Performing a metasearch (ie a search engine that is actually an aggregator of results from the real search engines), however, could assist in alleviating this problem. Other similar approaches would include the work done by OKI with their repository OSID. Something like their approach could provide the interoperability at either search engine or repository level that could also provide value. Users can also select search targets from those available to them or they could add new ones.