What search engines will be in the future will partly depend on the shortcomings of present engines DoingStuff.DonaldNoyes.20141113
There are many different search engines to process pages based on queries presented to them. They return many pages which satisfy the specific nature of the query, but also return many pages which do not satisfy the intent of the query.
It cannot be said then that the engines return "correct" responses because so many, sometimes a major portion of them, do not satisfy the searcher's intent. They do discover many "matches", but do not discover the context of the document or sometimes ignore important things such as punctuation. They often find matches spread apart, separated by commas, periods, quotation marks, in reverse order, and so on.
The SearchEngineOfTheFuture will
- consider these important considerations in addition to the matching of the words in the query. They will also consider the fact that some expressions in the query may be matched by content worded slightly differently, but having the same meaning. They will also consider the context of the location of the match, and the relative importance of it in the context. If the document is about cooking biscuits, and the query is about programming concepts, the context will be considered more than the exactness of the match.
- rank the matching documents in more than one area of consideration. If the document is of the correct type and the match is exact in two cases, but the author of one is deemed more expert in the field, it will receive a higher ranking. If the document has been subjected to peer review it might be ranked higher than one which is not.
- consider the presence of other related material in the document as a factor in raising the potential ranking of the page over one in which the match is more isolated and less supported.
- sometimes return few positives, so the rigor of comparison which might be applied should become somewhat more relaxed,
- when presented with thousands and millions of matches, exact more and more constraints to be applied, and the presentation of results might be accompanied by a grouping statement which describes the document type, time written, number of other documents which have been generated by its author, and so on.
- in the case of many matches, suggest clarification of the query so as to zero in on the intent which is not obvious or specific enough.
- will be expected to run on the very finest computers which are not only very fast, but are also efficient, and which accumulate results (learn) based on each search.
- will also have something new. It will work with other search engines, collaboratively identifying and classifying the pages it discovers by tagging the page in a separate but available space, not unlike a library card catalog.
- will include a place where when a query scores a hit a number will be assigned identifying tag ranking curiosity or relevance. This would be one of a family of identifications or classifications made by the engine, which might number into scores, or hundreds, or more.
Correct Responses?
It cannot be said that these SearchEngines return "correct" responses because so many, sometimes a major portion of them, do not satisfy the Searcher's intent. They may discover many "matches", but do not discover the context of the document or sometimes ignore important things such as punctuation. They often find matches spread apart, separated by commas, periods, quotation marks, in reverse order, and so on.
Considerations included in a Container
The SearchEngineOfTheFuture will consider these important considerations in addition to the matching of words or meeting certain conditions or accessorizations in the InitiatingContainer?. They will also consider the fact that some words in this container may be matched by existing words which mean the same or nearly the same thing. They will also consider such things as content, source and time or origin, derivative sources utilized in generating the context in which the match or fulfillment exists. Also to be considered is the relative importance that exists in the action and comparison.
For example, if the located document is primarily about cooking biscuits and the InitiatingContainer? is about programming concepts, the context will be considered as a factor in any ranking of the match or mismatch of the discovery.
The SearchEngineOfTheFuture
- in the case of many matches, suggest clarification of the query so as to zero in on the intent which is not obvious or specific enough.
- can be expected to run on the very finest computers which are not only very fast, but are also efficient, and which accumulate results (learn) based on each search.
- will also have something new. It will work with other engines, collaboratively identifying and classifying the pages it discovers by tagging the page in a separate but available space, not unlike a library card catalog. This space will be included in future situations where a query scores a hit. The number of such hits raising an identifying tag ranking curiosity or relevance. This would be only one of the identifications or classifications on the card, there might be scores and hundreds more.
- might find concepts present in TopicMaps (XTM) [XML Topic Maps] fitting to use, since such a map might be created for the query and for the matching page. This will make the collaborative document much more standardized and flexible at the same time than an informal set of classifications, categorizations and indexes.
- will include of any number of different topic maps can be created for a single set of resources. Furthermore, topic maps can be merged, which means that responsibility for developing a topic map can be delegated to several different people." (or agents) -- ProfessionalXmlMetaData Wrox Press 2001, ISBN 1861004516 Chapter 7 (quote from pg. 251)
- will include all of the results that match; they will have complete coverage.
- will align related searches of different people to deepen results.
- will get these people in touch with each other so that they can collaborate.
Design Issues
Related
CategorySemanticWeb CategoryFuture
Search Engine Of The Future