Records in this category

Sticky FAQs


ID #1044

What exactly is getting searched when Keywords is searched vs. Description vs. Subject?

Some people claim that federated search is a "dumbing down," or "the lowest common denominator" of searching. There is some truth to this in that if you're searching across databases and indexes there has to be something common to search on, or you just have a failure. Also, there may be some "dumbing" in that you cannot search across several databases for an index that is present only in one of the databases.

That said, Muse technology allows us to return results somewhere between "dumbed down" and what you would get using all the unique tools of the native search engine. It is always necessary and valuable to compare the results of searches using Muse technology to searches directly to the target database. However, like any comparison, the results are only valuable when the beginning variables are comparable (when apples are compared to apples). When we map Muse queries to the native target we do so using the default parameters of the native advanced search screen, so you should make comparisons based on those parameters without invoking any others. Additionally, we do not typically map wildcard search characters, and we only parse queries for Author, Title, Subject, Description, and Keyword (where applicable). These are the default characteristics of our Source Packages.

We can do much more specific mappings and parsings than this, but we do not do so as a matter of course. When we fail to obtain results similar or equal to the native search within the conditions noted above, it is reasonable to assume we have something wrong or incomplete in our Source Package mapping or parsing, and we can repair this when it is pointed out. Keep in mind that any given target will map a "description" index in different ways—that is, they will not all include the same fields in this index. We've seen everything from only indexing the 245 field in USMARC records to an index of every field in a full text record. Exactly what is indexed in the native records is not our choice—it is the particular vendor’s choice. So, since different vendors define indexes differently, the results of queries will vary from one vendor to another, and there is no definitive answer as to what exactly is getting searched when "Keyword" is searched vs. "Description" vs. "Subject" in a cross-database searching exercise.

If results vary between a Muse search and a native search where only defaults are used, then we should investigate the Muse Source Package to be certain it is parsing queries and interpreting results correctly. In other words, a “keyword” search at one source is not the same as a “keyword” search at another. Furthermore, differences will not necessarily be evident by scrutinizing the Muse Source Package Mapping or query parsing. The difference can be in the way the native indexes are built and the native search engine uses them. This is important because Muse technology provides neither an indexing nor a search engine. Muse queries use both the native indexes and the native search engine under the native default choices in the native “advanced” search screen That is why the most appropriate comparison would be between what Muse returns and what the native source returns for a search using the native search defaults in the advanced mode.

Federated searching is not intended to take the place of precise searching in any set of indexes or database. As search syntax gets more precise or specific to a given vendor or database, it will it will be less applicable to other vendors or databases. Federated search applications can mitigate that difference to a degree. But differences in targets imply a degree of difference that would not be obtained if you were searching one consistent set of indexes. It is useful to think of the federated search tool as a discovery engine comparable to Web search engines, except that Muse is capable of including subscribed and other sources not available on the open Web. On the open Web, if you know the site you want and how to get there, you’re better served to go directly to that site rather than to go through a Web search engine. The same is true of federated searching—if you know just where and how to find exactly what you need in a specific database, it’s better to search that database than to go through MultiSearch. But, just like the Web, if you always go to the same site for your information, you won’t necessarily find out about information published by other sites.

Tags: attributes, descripton, index, keyword, search, subject, type

Related entries: -

Last update: 2013-09-02 11:29
Author: Administrator
Revision: 1.0

Print this record Send FAQ to a friend Show this as PDF file
Rate this FAQ

Average rating: 0 (0 Votes)

completely useless 1 2 3 4 5 most valuable

You cannot comment on this entry

powered by phpMyFAQ 2.7.2