Ranked Search

Ranked search adds another level of complexity to search.

In looking at the search process, there are 5 main functions: Data Collection, Indexing, Query Processing, Search Engine, and Results Processing.

The Data Collection process is comprised of the process of collecting and complying the data.  This can be done with web crawlers, file traversers, and database connectors. 

The data is then converted, processed and tokenized as part of the Indexing process with such steps as lemmatization, entity extraction, and taxonomy classification, among other processes.  This data is then indexed and stored.

The Query Processing function consists of taking the user’s query and transforming the query.  The query processing transformation can involve such steps as checking for spell errors, automatic language detection, typos, official variants, stemming, entity extraction, phrasing, anti-phrasing, and stop words, among other checks. The result is a refined query that is ready to be processed by the search engine.

When a refined query arrives at the search engine, it takes the refined query and searches the stored index.  When it finds matches between the refined query and the data in the index, it outputs those raw results.

Once the search engine outputs the raw results, there is the final step of Results Processing.  This step might include filtering the results, removing duplicate results, merging the results from different sources, and then processing the results by sorting, ranking and paging the search results.  The user then receives the search set.

With the horsepower of Perfect Search, more time is available to employ more sophisticated ranking algorithms and achieve more relevant results.