Document retrieval over networks wherein ranking and relevance scores are computed at the client for multiple database documents

• US 5,659,732 A
• Filed: 05/17/1995
• Issued: 08/19/1997
• Est. Priority Date: 05/17/1995
• Status: Expired due to Term

• Current Assignee: Google Inc. (Alphabet Inc.)
• Original Assignee: Infoseek Corporation (The Walt Disney Company)
• Inventors: Kirsch, Steven T.
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): Rones, Charles L.
• Application Number:
  US08/443,348
• Time in Patent Office: 825 Days
• Field of Search: 364/419.9, 395/600, 395/609, 395/610, 395/605, 395/604
• US Class Current: 1/1
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9538 Presentation of query results
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99939 Privileged access

Abstract

A document search method using a plurality of databases available from one or more servers using one or more search engines. For each database, the number of records is determined and reported, as well as frequency of search query term occurances or hits, together with identification of database records corresponding to the hits. Reports from a plurality of databases are furnished to a user terminal, a client, where client software computes a relevance score for each record based upon the number of records in the database, the number of records having at least one hit and the number of hits for each record. This local computation from uniform data allows all documents to be ranked consistently as if coming from a single database.

Claims

1. A method for searching a plurality of databases which are distributed and accessible to a client through one or more search servers comprising,

(a) applying a search query from the client to each server associated with each database,

(b) determining at each server a list of relevant documents,

(c) obtaining, at the client from each server, statistics about each database,

(d) obtaining, at the client from each server, information about the relevant documents resulting from application of the query to the associated database, and

(e) computing, at the client, a relevance score for each document using said statistics and said information whereby the computed relevance score is used in determining how the relevant documents from all of the databases should be ordered in a list of merged relevant documents.

2. The method of claim 1 wherein the statistics about said collection include the size of the collection in terms of the number of records.

3. The method of claim 1 wherein the information about each document includes the number of times each search term appeared in the document.

4. The method of claim 1 wherein the information about each database includes the number of documents containing each search term.

5. A method for searching a plurality of databases which are distributed and accessible to a client through one or more servers comprising,

(a) accessing each database from the client,

(b) applying a search query from the client to the server associated with each database,

(c) obtaining, at the client, statistics about each database,

(d) obtaining, at the client, statistical information about the relevant documents resulting from application of the query to the database, including term count information for each document, and

(e) computing, at the client, a global score for the relevant documents in a list of merged relevant documents from all of the databases using said statistics and said information whereby the computed relevance score is used in determining how the relevant documents from all of the databases should be ordered in a list of merged relevant documents.

6. A method for searching text documents among a plurality of databases accessible through a server, in response to a search query at a client,

(a) applying a search query at the client to each database accessed through a server,

(b) for each database determining the total number of documents,

(c) for each of said databases, determining at the server the number of hits of each search query term for each relevant database document,

(d) for each of said databases, and for each query term, counting the documents having at least one hit, and

(e) computing at the client a relevance score of each record with respect to the search query, using the results of steps (b), (c) and (d) whereby the computed relevance score is used in determining how the relevant documents from all of the databases should be ordered in a list of merged relevant documents.

7. The method of claim 6 further defined by selecting a number of databases, having more than one search engine for the databases, prior to applying the same search query to all databases.

8. The method of claim 7 by selecting a number of records to be reviewed from among said number of databases, said number having the highest relevance scores for the search query.

9. The method of claim 8 further defined by preselecting a number of records prior to computing a relevance score.

10. A method for searching a plurality of databases which are distributed and accessible to a client through one or more search servers comprising,

providing a plurality of search engines each associated with a server storing a database,

applying a search query from the client to each search engine associated with a server storing a database, the search engines determining at each server a list of relevant documents,

obtaining, at the client from each server, statistical information about the relevant documents resulting from application of the query to the associated database, and

computing, at the client, on a single pass, a relevance score for each document using said statistical information about the relevant documents whereby the computed relevance score is used in determining how the relevant documents from all of the databases should be ordered in a list of merged relevant documents.

11. A method for searching a plurality of diverse databases which are distributed over a network and accessible to a client through one or more search servers using a plurality of search engines each associated with a server storing a database comprising,

obtaining, at the client from each search server, statistical information about relevant documents resulting from application of a query to a database associated with a search engine, and

computing, at the client, on a single pass, a relevance score for each document using said statistical information whereby the computed relevance score is used in determining how the relevant documents from diverse databases should be ordered in a list of merged relevant documents.