A Taxonomy of Web Search

Posted on


It is often regarded that the requirement behind a search query is to gather information, therefore classification of web queries can separated into 3 segmented classes:

The Navigation class is defined as the purpose to reach a site required.

The Information class is defined as the requirement to obtain information on the sites many web pages.

The Transaction class is defined as the intention to perform an activity guided by the site.

Navigation class queries – where a user performs a search for the purpose of reaching a site they have visited before or intend to reach a site that they assume to exist. Examples of these queries include:

Searoad Ferries. Intent to reach ‘http://www.searoad.com.au’

AMF bowling. Intent to reach ‘http://www.amfbowling.com.au’

Frank Oneil. Intent to reach ‘http://svpartners.com.au/our-people/frank-oneill/’

These search queries in information retrieval are also called keywords and is mostly used to evaluate results returned. The Text Retrieval Conference concluded that based on 150 queries against the WT10g collection distributed by CSIRO, these search queries are formally known as navigation class queries.

Information class queries – where a user performs a search to find information they assume will be on the internet in a non-variable format.  This means the expectation is to find a readable document that is static and requires no other user interaction, although dynamic pages are becoming more paramount to recent generation of search engines. Furthermore, information class queries can bring up a wider scale of results, for example boats or New York.

Transaction class queries – where a user performs a search and expects to reach a website that requires further interaction.  The interaction involved makes up the transaction defined in the class queries.  Examples consist of shopping, file sharing, finding database heavy services; results are generally difficult to evaluate, for instance a search query for interstate removalists Hobart will attempt to return results based on whether it is appropriate or not, other factors important to users such as price of products and efficiency of service are unavailable until a degree of interaction is performed.

How Taxonomy Can Improve Search

Posted on

Search engines have become a necessary tool to assist users in searching for relevant information not only on the web but also within large multi-national organizations that are content intensive.  Although technologies can change and different search methods are introduced, search engine taxonomy can be vital in improving and enhancing how content is being found.  Taxonomies can typically further improve the search quality by enhancing accuracy and returning user desired results. Let’s have a look at how taxonomy can work to improve search.

Control vocabulary

Content is indexed by what is known as spiders or web crawlers and stored in a database for retrieval when a string or search query matches words, phrase or text within the documents.  The problem with matching text is that words can be worded in many different ways to alter the context as well as words can have several definitions. A taxonomy would be able to associate the phrases and words that are relevant to the search.

Use of metadata

Although not all content is structured and contains relevant metadata such as title, author, description, subject, audience, etc.; for the content that do contain metadata information, it is important to take advantage and leverage from this metadata for use with controlled vocabularies.

Hierarchy structures

Hierarchical structures are a classic example of taxonomies and terms can be arranged in the order of relevance such as parent and child.  By creating a structure of terms, users will be able to search for the most desired term even if they do not know the correct terminology for a subject in mind.

Suggest search queries

It is not uncommon for users to perform search queries on a subject that they do not know how to define or do not known what they are searching until they see it.  A taxonomy can suggest search queries relating to a user’s set of strings and make a relationship with content that may be of interest.