System to Model and Query Relational Database and Text Index as One Search Interface
Publication Date: 2010-Sep-09
The IP.com Prior Art Database
Disclosed is a system which provides a webservice abstraction to model and query both text index and relational database as one datasource. This Unified Query API enables simple client applications to provide a complete faceted search user experience over all fields in either source, while leveraging the speeds of the text index for indexed fields.
System to Model and Query Relational Database and Text Index as One Search
Current state-of the-art text indices allow very quick faceted search and document
They provide an application protocol interface (API) to:
1. Search selected fields of a vast collection of documents
2. Retrieve large document streams sorted by a selected field, where each
document contains the selected fields
3. Aggregate counts over values of some selected fields of the documents where
each list of selected fields (search, retrieve, sort, aggregate) is provided at query
time by the client.
These indices query much more quickly than relational databases, which can also
search, retrieve, and aggregate, although more slowly. However, relational databases
outperform indices in load and update times, and it is impractical to store very large
fields or index frequently updating fields in the index.
The problem is that most search applications can only interface with one source and
have to choose between using a relational database which provides slow response
times, and using a text index which can't retrieve very large fields or search/aggregate
frequently updated fields.
No known systems attempt to leverage both the disparate advantages of the two
datasources. The closest solutions focus on adding functionality to the individual
Specifically, database vendors have created solutions to enable a text search inside the
database. This only helps search speed, and does not accelerate the document
retrieval or aggregation capabilities of a relational database.
Text indices are optimizing update times, and could use compression to enable storing
large text fields. No known efforts exist which enable faster updating of a field for many
documents in a search index.
An additional solution is required that enables leveraging the advantages of both types
of datasources. No known system exists to enable querying both relational database
and text index in an efficient manner.
Search applications typically cannot Interface with both text index and relational
database, as that dramatically increases application complexity.
The invented system provides a webservice abstraction to model and query both the
text index and the relational database as one datasource. This Unified Query API
enables simple client applications to provide a complete faceted search user experience
over all fields in either source, while leveraging the speeds of the text index for indexed
Creating this abstraction without this webservice requires an overly-complex client
Through the interface, clients provide a search query and retrieve documents or
aggregate counts. However, the fields referenced can exist in the database, the index,