|
||
Meta search engines can help find results you might otherwise miss. However, combining results from many engines into a single list brings up some issues. By understanding the problems and their solutions, you can choose a better meta search engine to use.
Problems that Affect Meta Search Engine Rankings
1) How to judge the importance of results? If there was broad agreement between the major search engines, ordering search results would be straightforward. Unfortunately, numerous studies have shown there is surprisingly little overlap among the top results of the major search engines.
In hindsight this is not surprising since there is enormous pressure to get high rankings using dubious tactics such as link farming or hidden content only seen by search engines. The result is a lot of “noise” in rankings when compared over different search engines.
2) How to pinpoint areas of interest? Many search engines offer suggestions for search terms that help focus on your area of interest. Eventually though, you have to look through lists of results. If meta search engines can help pinpoint interesting results in the list through indexing, clustering, or re-ranking, that saves your valuable time.
Ranking Methods used by Meta Search Engines
1) Original ranking – one method is just to use the original rankings returned by the search engines. Some adjustment is necessary to break ties or boost results perceived to be more relevant. For example, dogpile.com changes the original ranking based on what search terms are used.
* Problematic – different search engine ranking methods are probably not compatible.
2) Vote ranking – an interesting method of ranking is to order by votes, where a web page gets a vote each time it is in the top 10 results of a search engine. Some meta search engines, such as Ixquick.com show you how many votes each result got.
* Good – however not many web pages get multiple votes.
3) Filtered ranking – this method is based on whether results contain certain words or phrases. You can choose words or phrases that filter out results or keep certain results with desirable phrases. Many desktop-based meta search engines have this feature.
* Useful but brittle - a result may be ruled in or out by accident depending on what specific word the author used.
4) Cluster ranking – once you have settled on good search words or phrases, eventually you must look through a list of results and pick out the most relevant ones. Clustering puts results into similar groups so that instead of looking through one long list, you immediately rule out non-relevant groups and concentrate on smaller lists of relevant groups. Clusty.com uses clustering to group results into smaller lists that are easier to search.
* Good – depending on the size and quality of the clusters.
5) Relevance ranking – Besides looking at the content of the search results, the latest meta search engines, such as iMetaSearch, can also look at which results you have marked as relevant and sort results based on your feedback. These meta search engines can re-order the results or clusters to automatically find results and clusters similar to the one you have marked.
* Good – user feedback is taken into account and all words plus their context are used to find similar results, not just a few specific words.