Search the Community
Showing results for tags 'faceted'.
Found 1 result
A previously created thread on this topic was closed in the meantime. I would suspect that the reason for this was that the yield of such a search has not been recognized or appreciated sufficiently. A faceted search is characterised by the fact that the user gradually reduces the number of search results by adding keywords that have been used in the results. For this purpose, the system displays the keywords whose inclusion would not reduce the result set to an empty quantity. Such a procedure proves to be extremely effective in discriminating against keyword registers because the user does not have to make any assumptions about the keywords actually used, since these are provided by the system, depending on the already reduced result set. In this way, the relevant entities can usually be limited to a manageable number of hits with just a few clicks, even in very large databases. At the same time, the procedure is very simple in technical implementation. It simply consists of repeating the following steps: The number of each keyword in the observed data volume is determined and displayed. The user selects one of these keywords. Taking this into account reduces the amount of data considered. Then the number of entries per keyword has to be determined again, i. e. step 1 has to be executed again. This type of successive filtering of the data makes the yield from the information value of the allocated keywords actually accessible. Individually applied, the value of a keyword must decrease to the extent that it is applied. The more often a keyword is used, the less value it has until it becomes completely useless. The procedure is inexpensive and extremely effective. Without such a filtering method, however, systematic keywording is of no practical value. The value of keywords is only determined by their combination, to the extent that the volume of data increases. I would recommend that you re-evaluate this function. The effectiveness of this discrimination procedure is evident. In Apache Solr and Lucene there are concrete implementations which make the process tangible.