US5826261A
Expired / Lapsed
US
⏱️ 0.0 years remaining
System and method for querying multiple, distributed databases by selective sharing of local relative significance information for terms related to the query
Assignee: Excite Inc
•
Filing: 1996-05-10
•
Granted: 1998-10-20
•
Expires: 2016-05-10
Abstract
Abstract A system, method, and various software products provide improved information retrieval performance from multiple document databases by retrieving from the multiple document databases in response to a user query, a set of documents that globally satisfy the query, even though each database maintains independent document indices, term frequency information, and scoring functions. The global search result approximates, to any desired degree of error, the search results that would have been obtained had the multiple document databases been globally indexed. This is done by sharing at the time the query is executed, a small subset of information about the local relative significance of terms related to the user's query, and from this information, determining a global relative significance of such terms. From the global relative significance, the individual document databases determine their query results, which are then merged into a global set of documents satisfying the query. The shared local relative significance information may be the inverse document frequency of each of a number of terms related to the query, or it may be the total frequency of each of such terms. The global relative significance may correspondingly be a global inverse document frequency, or a global term frequency from which the global inverse document frequency is calculated.
Inventors
- 👤 Graham Spencer
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-26 22:45
CPC Classifications
G; G06; G06F; G06F16/00; G06F16/30; G06F16/31; G06F16/313; G06F16/33; G06F16/3331; G06F16/334; Y; Y10; Y10S; Y10S707/00; Y10S707/99931; Y10S707/99932; Y10S707/99933; Y10S707/99935
Claims Analysis
Total Claims: 46
Claim 1
Independent Claim
193 words
In an information retrieval apparatus including a database of documents, each document having a plurality of terms and a unique document indicia, the information retrieval apparatus further including a programmed processor adapted to receive a query containing at least one term and to compute in response to the query a document score for each of a selected plurality of documents, the document score being a function of the terms of the query, a computer memory readable by the processor and comprising: a first table that specifies for each document in the database a set of terms, each term in the set associated with a scalar measure of a contribution the term makes to the document score of the document, the scalar measure of contribution for a term being a function of an inverse document frequency of the term, and a frequency of the term within the document, the terms in each set ordered by the scalar measure of contribution, the sets ordered with respect to indicia of the documents, such that the information retrieval apparatus can determine for any document a selected number of terms that most significantly contribute to the document score.
Claim 2
Dependent Claim
193 words
In an information retrieval apparatus including a database of documents, each document having a plurality of terms and a unique document indicia, the information retrieval apparatus further including a programmed processor adapted to receive a query containing at least one term and to compute in response to the query a document score for each of a selected plurality of documents, the document score being a function of the terms of the query, a computer memory readable by the processor and comprising: a first table that specifies for each document in the database a set of terms, each term in the set associated with a scalar measure of a contribution the term makes to the document score of the document, the scalar measure of contribution for a term being a function of an inverse document frequency of the term, and a frequency of the term within the document, the terms in each set ordered by the scalar measure of contribution, the sets ordered with respect to indicia of the documents, such that the information retrieval apparatus can determine for any document a selected number of terms that most significantly contribute to the document score.
Claim 3
Dependent Claim
98 words
The computer readable memory of claim 1 wherein the error contribution weight for a term T is determined by the formula: ##EQU9## where: D iterates over a set of documents containing the term; W T is a weighting factoring for term T; IDF T is an inverse document frequency value determined as a function of the frequency of term T in the database; TF D ,T is a within-document frequency value determined as a function of the frequency of term T in document D; and, k D is a normalization factor for a document vector of document D.
Claim 4
Dependent Claim
217 words
In an information retrieval apparatus including a database of documents, each document having a plurality of terms and a unique document indicia, the information retrieval apparatus further including a programmed processor adapted to receive a query containing at least one term and to compute in response to the query a document score for each of a selected plurality of documents, the document score being a function of the terms of the query, a computer memory readable by the processor and comprising: a first table that specifies for each document in the database a set of terms, each term in the set associated with a frequency of the term in the document, the terms in each set ordered by a scalar measure of a contribution that each term makes to the document score of the document, the scalar measure of contribution for a term being a function of an inverse document frequency of the term, and a frequency of the term within the document, the sets ordered with respect to indicia of the documents, such that the information retrieval apparatus can determine for any document a selected number of terms that most significantly contribute to the document score by retrieving frequency information from the set of terms for the document and computing the contribution from the frequency information.
Claim 5
Dependent Claim
217 words
In an information retrieval apparatus including a database of documents, each document having a plurality of terms and a unique document indicia, the information retrieval apparatus further including a programmed processor adapted to receive a query containing at least one term and to compute in response to the query a document score for each of a selected plurality of documents, the document score being a function of the terms of the query, a computer memory readable by the processor and comprising: a first table that specifies for each document in the database a set of terms, each term in the set associated with a frequency of the term in the document, the terms in each set ordered by a scalar measure of a contribution that each term makes to the document score of the document, the scalar measure of contribution for a term being a function of an inverse document frequency of the term, and a frequency of the term within the document, the sets ordered with respect to indicia of the documents, such that the information retrieval apparatus can determine for any document a selected number of terms that most significantly contribute to the document score by retrieving frequency information from the set of terms for the document and computing the contribution from the frequency information.
Claim 6
Dependent Claim
98 words
The computer readable memory of claim 3 wherein the error contribution weight for a term T is determined by the formula: ##EQU10## where: D iterates over a set of documents containing the term; W T is a weighting factoring for term T; IDF T is an inverse document frequency value determined as a function of the frequency of term T in the database; TF D ,T is a within-document frequency value determined as a function of the frequency of term T in document D; and, k D is a normalization factor for a document vector of document D.
Claim 7
Dependent Claim
208 words
In a computer that is communicatively coupled to a plurality of databases, each database maintaining a set of documents arranged and indexed independently of the other databases, each term having a local relative significance within each database that is independent of the local relative significance of the term in the other databases, a computer memory readable by a processing device of the computer and configuring and controlling the processing device to perform the steps of: requesting from each of the databases for each of a first list of terms, including terms of the query, the local relative significance of the term in the database; combining the first lists of terms from the databases, and determining for each of the terms a global relative significance of the term from the local relative significance of the term in each of the databases; receiving from each of the databases a local search result of documents locally satisfying the query executed on the database using the determined global relative significance of selected terms in place of the local relative significance of the selected terms in that database; and combining the local search results into a global search to produce the list of documents satisfying the query with respect to the multiple databases.
Claim 8
Dependent Claim
208 words
In a computer that is communicatively coupled to a plurality of databases, each database maintaining a set of documents arranged and indexed independently of the other databases, each term having a local relative significance within each database that is independent of the local relative significance of the term in the other databases, a computer memory readable by a processing device of the computer and configuring and controlling the processing device to perform the steps of: requesting from each of the databases for each of a first list of terms, including terms of the query, the local relative significance of the term in the database; combining the first lists of terms from the databases, and determining for each of the terms a global relative significance of the term from the local relative significance of the term in each of the databases; receiving from each of the databases a local search result of documents locally satisfying the query executed on the database using the determined global relative significance of selected terms in place of the local relative significance of the selected terms in that database; and combining the local search results into a global search to produce the list of documents satisfying the query with respect to the multiple databases.
Claim 9
Dependent Claim
212 words
A computer implemented method of querying multiple databases with a query having at least one term to produce a list of documents from all of the databases that satisfy the query, each term having a total frequency in each database, each database having a total number of documents and an independently determined inverse document frequency value (IDF) for each unique term in the database, the IDF being a function of a number of documents in the database including the term, the method comprising: requesting from each of the databases the total frequency of each of a first list of terms, and determining therefrom a global frequency for each of the first list of terms; determining a global number of documents in all of the databases from the total number of documents in each database; in each database computing a global IDF for each term of the query from the global frequency of each term, and the global number of documents, and executing the query to produce a list of documents in the database that satisfy the query using the global IDFs of the terms of the query; and, merging the lists of documents from the database to produce a list of documents globally satisfying the query with respect to the multiple databases.
Claim 10
Dependent Claim
212 words
A computer implemented method of querying multiple databases with a query having at least one term to produce a list of documents from all of the databases that satisfy the query, each term having a total frequency in each database, each database having a total number of documents and an independently determined inverse document frequency value (IDF) for each unique term in the database, the IDF being a function of a number of documents in the database including the term, the method comprising: requesting from each of the databases the total frequency of each of a first list of terms, and determining therefrom a global frequency for each of the first list of terms; determining a global number of documents in all of the databases from the total number of documents in each database; in each database computing a global IDF for each term of the query from the global frequency of each term, and the global number of documents, and executing the query to produce a list of documents in the database that satisfy the query using the global IDFs of the terms of the query; and, merging the lists of documents from the database to produce a list of documents globally satisfying the query with respect to the multiple databases.
Claim 11
Dependent Claim
202 words
In a computer that is communicatively coupled to a plurality of databases, each database maintaining a set of documents arranged and indexed independently of the other databases, each term having a local relative significance within each database that is independent of the local relative significance of the term in the other databases, a computer memory readable by a processing device of the computer and configuring and controlling the processing device to perform the steps of: requesting from each of the databases the local IDF of each of a first list of terms, and determining therefrom a global IDF for each of the first list of terms, the global IDF of a term being a function of a total number of documents in all of the databases, and total number of documents containing the term; receiving from each database a list of documents in the database that satisfy the query executed on the database while substituting the global IDFs of at least the terms of the query for the local IDFs of the terms in the database; and, merging the lists of documents from the database to produce a consolidated list of documents globally satisfying the query with respect to the multiple databases.
Claim 12
Dependent Claim
202 words
In a computer that is communicatively coupled to a plurality of databases, each database maintaining a set of documents arranged and indexed independently of the other databases, each term having a local relative significance within each database that is independent of the local relative significance of the term in the other databases, a computer memory readable by a processing device of the computer and configuring and controlling the processing device to perform the steps of: requesting from each of the databases the local IDF of each of a first list of terms, and determining therefrom a global IDF for each of the first list of terms, the global IDF of a term being a function of a total number of documents in all of the databases, and total number of documents containing the term; receiving from each database a list of documents in the database that satisfy the query executed on the database while substituting the global IDFs of at least the terms of the query for the local IDFs of the terms in the database; and, merging the lists of documents from the database to produce a consolidated list of documents globally satisfying the query with respect to the multiple databases.
Claim 13
Dependent Claim
217 words
A computer implemented method of querying multiple databases with a query having at least one term to produce a list of documents from all of the databases that satisfy the query, each database maintaining a set of documents arranged and indexed independently of the other databases, each term having a local relative significance within each database that is independent of the local relative significance of the term in the other databases, the method comprising, requesting from each of the databases for each of a first list of terms, including terms of the query, the local relative significance of the term in the database; combining the first lists of terms from the databases, and determining for each the terms, a global relative significance of the term from the local relative significance of the term in each of the databases; in each database, after determining the global relative significance of each of the terms, executing the query on the database using the global relative significance of selected terms in place of the local relative significance of the selected terms in that database, to produce a local search result of documents locally satisfying the query; combining the local search results into a global search result to produce the list of documents globally satisfying the query with respect to the databases.
Claim 14
Dependent Claim
217 words
A computer implemented method of querying multiple databases with a query having at least one term to produce a list of documents from all of the databases that satisfy the query, each database maintaining a set of documents arranged and indexed independently of the other databases, each term having a local relative significance within each database that is independent of the local relative significance of the term in the other databases, the method comprising, requesting from each of the databases for each of a first list of terms, including terms of the query, the local relative significance of the term in the database; combining the first lists of terms from the databases, and determining for each the terms, a global relative significance of the term from the local relative significance of the term in each of the databases; in each database, after determining the global relative significance of each of the terms, executing the query on the database using the global relative significance of selected terms in place of the local relative significance of the selected terms in that database, to produce a local search result of documents locally satisfying the query; combining the local search results into a global search result to produce the list of documents globally satisfying the query with respect to the databases.
Claim 15
Dependent Claim
24 words
The computer implemented method of claim 8 further comprising a first step of: requesting from each of the databases the first list of terms.
Claim 16
Dependent Claim
102 words
The computer implemented method of claim 9, wherein requesting from each of the databases the first list of terms comprises for each database: providing the query to the database; executing the query to obtain a first set of documents locally satisfying the query; determining from the first set of documents a second list of terms that significantly influence the results of the query as a function of the frequency of the term in each of the first set of documents and the frequency of the term in the database; and, selecting the first list of terms from the second list of terms.
Claim 17
Dependent Claim
140 words
The computer implemented method of claim 10, wherein: determining from the first set of documents therein a second list of terms that significantly influence the results of the query comprises: accumulating across each of the first set of documents an error contribution weight for selected terms in the first set of documents, the error contribution weight being a function of an inverse document frequency of the term, and a frequency of the term within the document, such that the selected terms form the second list of terms; and, selecting the first list of terms from the second list of terms comprises selecting either: terms having an error contribution weight greater than a predetermined or adaptively determined threshold value; or, a predetermined or adaptively determined number of terms; or, a predetermined or adaptively determined percentage of the second list of terms.
Claim 18
Dependent Claim
102 words
The computer implemented method of claim 11 wherein the error contribution weight for the selected terms is determined by the formula: ##EQU11## where: T iterates of the selected terms; D iterates over the first set of documents; W T is a weighting factoring for term T; IDF T is an inverse document frequency value determined as a function of the frequency of term T in the database; TF D ,T is a within-document frequency value determined as a function of the frequency of term T in document D; and, k D is a normalization factor for a document vector of document D.
Claim 19
Dependent Claim
101 words
The computer implemented method of claim 9, wherein: each document is represented by a document vector having a vector length; and, requesting from each of the databases the first list of terms, comprises for each database: i) providing the query to the database; ii) executing the query to obtain a first set of documents locally satisfying the query; iii) selecting from the first set of documents a set of highly ranked documents; and, iv) selecting from the set of highly ranked documents terms that significantly contribute to the vector length of the document vector of at least one highly ranked document.
Claim 20
Dependent Claim
74 words
The computer implemented method of claim 8, wherein determining for each of the terms a global relative significance of the term from the local relative significance of the term in each of the databases comprises: determining the global relative significance of a term as a function of; a combined number of documents in all of the databases; and, a combined frequency of the term in all of the documents in all of the databases.
Claim 21
Dependent Claim
50 words
The computer implemented method of claim 14, wherein the global relative significance of the term, G T is determined by the formula: ##EQU12## where N i is the number of documents in database i, and n i ,T is the number of documents in database i that contain term T.
Claim 22
Dependent Claim
27 words
The computer implemented method of claim 8, wherein the global relative significance of the terms is determined by a client computer requesting the query from the databases.
Claim 23
Dependent Claim
48 words
The computer implemented method of claim 8, wherein the global relative significance of the terms is determined by the individual databases from a total frequency of each of the significant terms in each of the other databases, and a total number of documents in all of the databases.
Claim 24
Dependent Claim
46 words
The computer implemented method of claim 8, wherein the local relative significance of a term in a database is a first inverse document frequency value, and the global relative significance is a second inverse document frequency for the term with respect to all of the databases.
Claim 25
Dependent Claim
127 words
The computer implemented method of claim 18, wherein: each document has a determinable document vector having a vector length as a function of the inverse document frequency of selected terms in the document; and, executing the query on the database using the global relative significance of selected terms in place of the local relative significance of the selected terms in that database comprises: replacing the local relative significance of the selected terms with the global relative significance of the selected terms; for each document that contains at least one of the selected terms having a global relative significance, redetermining the length of the document vector of the document according to the global relative significance of the term; and, executing the query by scoring and then ranking documents.
Claim 26
Dependent Claim
31 words
The computer implemented method of claim 19, wherein the documents containing at least one of the terms having a global relative significance further contain at least one term of the query.
Claim 27
Dependent Claim
50 words
The computer implemented method of claim 8, wherein the local relative significance of a term in a database is a total frequency of the term in the database, and the global relative significance of the term is a global frequency of the term with respect to all of the databases.
Claim 28
Dependent Claim
153 words
The computer implemented method of claim 21, wherein: each document has a determinable document vector having a vector length as a function of an inverse document frequency of selected terms in the document; and, executing the query on the database using the global relative significance of selected terms in place of the local relative significance of the selected terms in that database comprises: for each selected term, replacing the total frequency of the term with its global frequency in all of the databases and computing a global inverse document frequency for the term from the global frequency and a total number of documents in databases; for each document that contains at least one of the selected terms having a global inverse document frequency, redetermining the length of the document vector of the document according to the global inverse document frequency of the term; and, executing the query by scoring and then ranking documents.
Claim 29
Dependent Claim
24 words
The computer implemented method of claim 8, wherein at least two of the databases are controlled by two computers physically remote from each other.
Claim 30
Dependent Claim
26 words
The computer implemented method of claim 8, wherein at least two of the databases are physically controlled by a single computer and maintain separate index information.
Claim 31
Dependent Claim
95 words
The computer implemented method of claim 8, further comprising: providing the query to a query router; determining with the query router a selected subset of the databases from which to retrieve documents satisfying the query; and, wherein said requesting from each of the databases for each of a first list of terms, including terms of the query, the local relative significance of the term in the database, comprises requesting only from the selected subset of databases, such that the resulting list of documents globally satisfies the query with respect to the selected subset of databases.
Claim 32
Dependent Claim
182 words
A computer implemented method of querying multiple databases with a query having at least one term to produce a list of documents from all of the databases that satisfy the query, each database having an independently determined inverse document frequency value (IDF) for each unique term in the database, the IDF being a function of a number of documents in the database including the term, the method comprising: requesting from each of the databases the local IDF of each of a first list of terms, and determining therefrom a global IDF for each of the first list of terms, the global IDF of a term being a function of a total number of documents in all of the databases, and total number of documents containing the term; in each database, executing the query to produce a list of documents in the database that satisfy the query accounting for the global IDFs of the terms of the query; and, merging the lists of documents from the database to produce a list of documents globally satisfying the query with respect to the multiple databases.
Claim 33
Dependent Claim
182 words
A computer implemented method of querying multiple databases with a query having at least one term to produce a list of documents from all of the databases that satisfy the query, each database having an independently determined inverse document frequency value (IDF) for each unique term in the database, the IDF being a function of a number of documents in the database including the term, the method comprising: requesting from each of the databases the local IDF of each of a first list of terms, and determining therefrom a global IDF for each of the first list of terms, the global IDF of a term being a function of a total number of documents in all of the databases, and total number of documents containing the term; in each database, executing the query to produce a list of documents in the database that satisfy the query accounting for the global IDFs of the terms of the query; and, merging the lists of documents from the database to produce a list of documents globally satisfying the query with respect to the multiple databases.
Claim 34
Dependent Claim
70 words
The computer implemented method of claim 26, wherein the first list of terms is determined by: in each of the databases, executing the query to produce a first set of documents, and selecting from the terms contained in the first set of document, terms for which the IDF contributes more than a threshold amount of error to a document score based on the IDFs of the terms in the document.
Claim 35
Dependent Claim
180 words
The computer implemented method of claim 26, wherein the first list of terms is determined by: in each of the databases, executing the query to produce a first set of documents, and determining for each of selected terms T in the first set of documents an error contribution weight E T : ##EQU13## where: D iterates over the first set of documents; W T is a weighting factoring for a term T; IDF T is an inverse document frequency value determined as a function of the frequency of term T in the database; TF D ,T is a within-document frequency value determined as a function of the frequency of term T in document D; k D is a normalization factor for a document vector of document D; and, selecting the first list of terms from the terms T as either: terms T having an error contribution weight E T greater than a predetermined or adaptively determined threshold value; or, a predetermined or adaptively determined number of the terms T; or, a predetermined or adaptively determined percentage of the terms T.
Claim 36
Dependent Claim
46 words
The computer implemented method of claim 26, wherein the global IDF, G T is determined by the formula: ##EQU14## where N i is the number of documents in database i, and n i ,T is the number of documents in database i that contain term T.
Claim 37
Dependent Claim
90 words
The computer implemented method of claim 26, wherein executing the query to produce a list of documents in the database that satisfy the query accounting for the global IDFs of the terms of the query comprises: for each document containing a term of the query for which there is a global IDF, recomputing a normalization factor of the document based on the global IDFs of the terms of the query; and computing a document score according to the IDFs of the terms of the query and the recomputed normalization factor.
Claim 38
Dependent Claim
199 words
A computer readable memory including an application programming interface for a database management system communicatively coupled to a client application and a database of documents, the memory storing: a first method invocable by the client application that receives from the client application a query containing at least one term, and returns to the client application a list of terms, each of the list of terms contributing to a document score of at least one document satisfying the query; a second method invocable by the client application that receives from the client application a second list of terms, and returns to the client application for each of the second terms a local relative significance of the term; a third method invocable by the client application that receives from the client application a query containing at least one term, and that returns to the client application a set of document identifiers of documents satisfying the query executed using a global relative significance of selected terms in place of the local relative significance of the selected terms, the global relative significance of each selected term computed from the local relative significance of the term in each of a plurality of document databases.
Claim 39
Dependent Claim
199 words
A computer readable memory including an application programming interface for a database management system communicatively coupled to a client application and a database of documents, the memory storing: a first method invocable by the client application that receives from the client application a query containing at least one term, and returns to the client application a list of terms, each of the list of terms contributing to a document score of at least one document satisfying the query; a second method invocable by the client application that receives from the client application a second list of terms, and returns to the client application for each of the second terms a local relative significance of the term; a third method invocable by the client application that receives from the client application a query containing at least one term, and that returns to the client application a set of document identifiers of documents satisfying the query executed using a global relative significance of selected terms in place of the local relative significance of the selected terms, the global relative significance of each selected term computed from the local relative significance of the term in each of a plurality of document databases.
Claim 40
Dependent Claim
86 words
The computer readable memory of claim 31, wherein the first method: i) executes the query to obtain a first set of documents locally satisfying the query; ii) determines from the first set of documents a second list of terms that significantly influence the results of the query as a function of the frequency of the term in each of the first set of documents and the frequency of the term in the database; and iii) selects the list of terms from the second list of terms.
Claim 41
Dependent Claim
89 words
The computer readable memory of claim 32, wherein the first method determines from the first set of documents a second list of terms that significantly influence the results of the query by: accumulating across each of the first set of documents an error contribution weight for selected terms in the first set of documents, the error contribution weight being a function of an inverse document frequency of the term, and a frequency of the term within the document, such that the selected terms form the second list of terms.
Claim 42
Dependent Claim
96 words
The computer readable memory of claim 33, wherein the error contribution weight for a term T is determined by the formula: ##EQU15## where: D iterates over the first set of documents; W T is a weighting factoring for term T; IDF T is an inverse document frequency value determined as a function of the frequency of term T in the database; TF D ,T is a within-document frequency value determined as a function of the frequency of term T in document D; and, k D is a normalization factor for a document vector of document D.
Claim 43
Dependent Claim
25 words
The computer readable memory of claim 31, wherein the third method further receives a list of terms, each term associated with a global relative significance.
Claim 44
Dependent Claim
26 words
The computer readable memory of claim 35 wherein the global relative significance is a global frequency of the term with respect to all of the databases.
Claim 45
Dependent Claim
28 words
The computer readable memory of claim 35 wherein the global relative significance is a global inverse document frequency of the term with respect to all of the databases.
Claim 46
Dependent Claim
59 words
The computer readable memory of claim 31, wherein the third method: scores each document that contains at least term of the query having a global relative significance by redetermining a length of the document vector of the document according to the global relative significance of the term, and computing a document score as a function the redetermined vector length.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2001-10-19 | 2002-04-20 | 2002-10-20 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2005-10-19 | 2006-04-20 | 2006-10-20 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2009-10-19 | 2010-04-20 | 2010-10-20 | $7700.00 | Window Closed / Lapsed |