Browse Prior Art Database

Publication Date: 2015-Jun-02
Document File: 5 page(s) / 125K

Publishing Venue

The Prior Art Database


A query suggestion system that automatically modifies an input query to a relevant and a meaningful query is disclosed. According to an embodiment, the input query is a query input by a service engineer to the query suggestion system that returns zero results. Such an input is taken in form of a SPARQL query. The query suggestion system uses an ontological structure along with a spell check module using character edit distance to generate alternate SPARQL queries. Such alternate SPARQL queries are presented as suggestions to the service engineer.

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 41% of the total text.



The present disclosure relates generally to search engine systems that enable query based searching of large document repositories and more particularly to a query suggestion system that enables automatic modification of a query to a relevant and meaningful query.

Generally, service engineers require frequent querying for different entities in service logs. The service engineers perform query based searching of large document repositories using one of the numerous search engines generally known in the art. The document repository includes engineering cases that include records of historic information on machine failures, components, corrective actions and other maintenance events. Such information about machine failure is in many instances captured in a form of free-flowing or unstructured text.  

Further, service logs include a large number of entities. Service engineers cannot be expected to know and remember name of every entity accurately. Purpose of the search engine is to  enable field service engineers retrieve data from the repository for quickly understanding and solving machine defects, by looking up similar cases in historic cases available in the repository. However, as service engineers do not remember names of all equipment and entities, typographical errors in query input by the service engineer are common and because of these errors the search engine fails to return results.

A conventional search engine system uses data of the service logs and a knowledge base with an underlying ontological structure. However, if an exact entity name is not searched for, as part of the query input, no result is returned. A lot of time may be spent by the service engineers in trying different queries without successfully resolving the issue. Figure 1 is a block diagram of architecture of the conventional search engine system.

Figure 1

The working of the conventional search engine system is described further. Initially, a query is issued in a form of a natural language query. A query interpreter module converts the query to a SPARQL Protocol and RDF Query Language (SPARQL) query. The SPARQL query is then issued to the knowledge base and results are returned to an output formatter. The output formatter presents the results to a user in a user friendly manner. However, if the user query includes spelling errors, the result count is zero.

Another conventional technique generates SPARQL queries from query in natural language to query an underlying ontology. In some conventional techniques, alternate query suggestions are generated. The alternate query suggestions may be context driven, provided to the user based on the query given by the user in a particular field or may be differently paraphrased alternatives of the user query. One conventional technique performs spell check to generate the alternate query suggestions. However, the convention...