Information-Access Services

At the application level, Verity K2 offers a variety of services that applications can use to give users functional and convenient access to the information that has been extracted, indexed, and classified by the information-extraction services described earlier.

Figure 1-2 shows some of the available information-access services. More information on each is available elsewhere in this book.

Search. Sophisticated search is at the core of the information services that K2 applications provide for users. The search capabilities include

 

Full-text search, supporting single or multiple words or phrases, case- or accent-sensitive or insensitive searches, synonyms, wildcards, proximity searches, and logical combinations of them.

Fuzzy searches, using automatic re-spelling or stemming of search terms.

Language-specific search, in any of a large number of languages.

Topic search, in which a single search term can expand into query expression of any size or complexity.

Federated Search. Federated search (licensed separately from K2 as Verity Federator) gives out-of-the-box access to many information sources, including proprietary sources licensed by Verity on behalf of its customers. Federator sends a single query to multiple sources such as internal Verity indexes plus external sources like Web sites and proprietary subscription sources such as news feeds and business information services. Federator then merges results from all of them into a unified presentation for the user.

 

Federator is a powerful, standalone application and API that combines federated search with the Verity Ultraseek technology to extend the reach of an enterprise’s search capabilities. See your Verity representative for more information.

Parametric Search and Taxonomy Browse. An application that works with information that has been processed into classification structures (parametric indexes and taxonomies) can use sophisticated information-access services such as the following:

 

Parametric selection, in which the user can find documents by selecting values for various parameters instead of by searching.

Text search combined with parametric selection, restricting the results to documents that match both the currently selected parameters and the search terms.

Taxonomy browse, in which users navigate through hierarchical categories of information to reach a desired set of results. The results at each stage consist of documents that match the selected taxonomy category, plus the currently selected parameters, plus any search terms that are applied.

Relational taxonomies, in which multiple taxonomies are applied to the data. The user can start browsing one taxonomy, then switch to another, and even back again, until arriving at a desired category or document. Results can be restricted to documents that simultaneously match the selected category in each of the represented taxonomies, plus the currently selected parameters, plus any search terms that are applied.

For example, a user might search for a used car by selecting values for categories such as color, mileage, price, and year, then navigate both geographic and manufacturer taxonomies, and finally search for a specific desired feature (such as a car alarm).



Universal Document Viewing. The Verity K2 viewing service provides applications with a powerful document viewing and highlighting service. When the user clicks a link on the search results page to view a document (in any of hundreds of supported formats), the viewing service displays the document content and highlights occurrences of the search terms throughout the document.

 

Spelling Suggestion. Spelling suggestion can be used to suggest corrections to mistyped words in a user’s query. If a search returns no or few results, the application can display a message on the search results page, listing a suggested alternate query. For example, if the user searches for “helo wonderful worlld” and that phrase returns no hits, your application could respond with

 

Are you searching for “hello wonderful world”?

See Providing Spelling Suggestion for more information.

Document Summarization. By presenting a short, automatically generated summary for each document in a results list, a K2 application can help users quickly assess the relevance of the returned documents before retrieving the documents themselves.

 

Verity K2 supports several kinds of document summaries, including passage-based summaries, which consist of text excerpts in which the search term appears, optionally highlighted. For more information about summarization, see Returning Document Summaries.

Document Clustering. When presenting search results to the user, an application can cluster, or group together, documents covering similar topics or concepts.

 

Clustering relies on the extraction of document features at collection-indexing time (see Extracting Document Features).

Recommendation. An application that uses the K2 Recommendation Engine can suggest or recommend documents, expert users, or other entities that are specifically relevant to the current user’s context. To do so, the Recommendation Engine uses profiles (recommendation indexes) that are continually updated with users’s actions and preferences, meaning that the recommendations can evolve over time.

 

Features available with the Recommendation Engine include

Adaptive ranking of search results

Context-based personalization in user profiles

Concept-based retrieval (independent of specific keywords)

Location of experts and communities

Session-based profiles (updated dynamically within a single session)

See Providing Recommendations for more information.