Browse Prior Art Database

Method and System for Localized Tagging and Sub-Categorization of Category Tree to improve Local Business Search Relevancy

IP.com Disclosure Number: IPCOM000199668D
Publication Date: 2010-Sep-14
Document File: 7 page(s) / 214K

Publishing Venue

The IP.com Prior Art Database

Related People

Saikat Debiprasad Kundu: INVENTOR [+2]

Abstract

A method and system for localized tagging and sub-categorization of a category tree to improve local business search relevancy is disclosed. The method performs localized sub-categorization and tagging so that localized terms can be used as category tags or sub-categories specific to localities that enhance the relevancy of the category tree.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 28% of the total text.

Method and System for Localized Tagging and Sub-Categorization of Category Tree to improve Local Business Search Relevancy

Abstract

A method and system for localized tagging and sub-categorization of a category tree to improve local business search relevancy is disclosed.  The method performs localized sub-categorization and tagging so that localized terms can be used as category tags or sub-categories specific to localities that enhance the relevancy of the category tree.

Description

Field Name

Value

Business Name

Birla Mandir

Business Address

Mandir marg, Gol market

Phone No

011 23363637

Category

Temple

City

New Delhi

State

Delhi

Figure 1:  Sample entity

Figure 2:  Category Tree

For instance, local entities such as business listings shown in Fig 1 are generally classified into a predefined hierarchical category tree as shown in Fig 2.  The top level of the category tree shows restaurant, entertainment and electronics as shown in Fig 2.  The restaurants are further classified into Chinese restaurants, north Indian restaurant, etc.  Sample entity with business name “Birla Mandir” shown in Fig 1 is classified under the category “Temple”.

Existing local search offerings return entities such as business listings shown in Fig 1 based on a user search query.  The keyword entered by the user is matched against the category names in category tree and entities belonging to the matched category are returned.  The user query “Temple” in “Delhi” is matched with the category “Temple” in the category tree and entities associated with the category are returned.  However, a search for “Mandir” (localized term for ‘Temple’ in North India) in “Delhi” may not return the entities under the general category “Temple” as the local term differs with the category name.  Thus, the local search will fail to return relevant results if the search query contains localized terms.


The disclosed invention uses the partially structured information contained in entity names (e.g. the term “Mandir” in entity name “Birla Mandir” shown in Fig 1) or from the user inputs of local intent.  Thereafter, the method sub-categorizes the associations with a locality associated with the existing category tree upon identifying sufficient support from a number of other entities in that locality.  However, if an association is a generic term across localities, then the method tags the association to the category.  These localized generic terms directly translate into a more relevant local search for the user as category words are given the highest weightage in most local queries.

In the first example, the localized term is assumed as “Mandir”.  The term “Mandir” is tagged with “Temples” for Mumbai locality.  Further another assumption is that the location or the locality intent of a user is known, based on the search query (‘Mumbai’ in this case) received as the input.  If the user searches for term “Mandir” in s...