Displaying implicit associations among items in loosely-structured data sets

Claim 1

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concepts and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities together as objects in said space.

Claim 2

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concepts and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities together as objects in said space.

Claim 3

A method for extracting inherent or implicit conceptual relationships or semantic associations that exist among items in a data set, and for representing said associations in a simulated three-dimensional space, said method comprising the steps of: organizing said items into a plurality of entity data pairs, each entity data pair comprising a label of an entity and a textual or symbolic entity description; creating a set of modified entity descriptions by pruning irrelevant terms from each textual or symbolic entity description and reducing each remaining term to a linguistic root form; extracting concepts from said set of modified entity descriptions, each concept comprising a root term associated with at least two modified entity descriptions; producing a similarity matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing a binary indication of whether the corresponding concept is present in the corresponding modified entity description; quantifying the associative structure of the data set by subjecting the similarity matrix to procedures comprising dual-scaling, in combination with a matrix transformation, thereby producing a set of coordinates for each concept and each entity in a multi-dimensional Euclidean space; and displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the multi-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 4

A method for extracting inherent or implicit conceptual relationships or semantic associations that exist among items in a data set, and for representing said associations in a simulated three-dimensional space, said method comprising the steps of: organizing said items into a plurality of entity data pairs, each entity data pair comprising a label of an entity and a textual or symbolic entity description; creating a set of modified entity descriptions by pruning irrelevant terms from each textual or symbolic entity description and reducing each remaining term to a linguistic root form; extracting concepts from said set of modified entity descriptions, each concept comprising a root term associated with at least two modified entity descriptions; producing a similarity matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing a binary indication of whether the corresponding concept is present in the corresponding modified entity description; quantifying the associative structure of the data set by subjecting the similarity matrix to procedures comprising dual-scaling, in combination with a matrix transformation, thereby producing a set of coordinates for each concept and each entity in a multi-dimensional Euclidean space; and displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the multi-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 9

An apparatus for extracting inherent or implicit conceptual relationships or semantic associations that exist among items in a data set, and for representing said associations in a simulated three-dimensional space, said apparatus comprising: means for organizing said items into a plurality of entity data pairs, each entity data pair comprising a label of an entity and a textual or symbolic entity description; means for creating a set of modified entity descriptions by pruning irrelevant terms from each textual or symbolic entity description and reducing each remaining term to a linguistic root form; means for extracting concepts from said set of modified entity descriptions, each concept comprising a root term that is associated with at least two modified entity descriptions; means for producing a similarity matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing a binary indication of whether the corresponding concept is present in the corresponding modified entity description; means for quantifying the associative structure of the data set by subjecting the similarity matrix to procedures comprising correspondence analysis, in combination with a matrix transformation operation, thereby producing a set of coordinates for each concept and each entity in a multi-dimensional Euclidean space; and means for displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the multi-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 10

An apparatus for extracting inherent or implicit conceptual relationships or semantic associations that exist among items in a data set, and for representing said associations in a simulated three-dimensional space, said apparatus comprising: means for organizing said items into a plurality of entity data pairs, each entity data pair comprising a label of an entity and a textual or symbolic entity description; means for creating a set of modified entity descriptions by pruning irrelevant terms from each textual or symbolic entity description and reducing each remaining term to a linguistic root form; means for extracting concepts from said set of modified entity descriptions, each concept comprising a root term that is associated with at least two modified entity descriptions; means for producing a similarity matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing a binary indication of whether the corresponding concept is present in the corresponding modified entity description; means for quantifying the associative structure of the data set by subjecting the similarity matrix to procedures comprising correspondence analysis, in combination with a matrix transformation operation, thereby producing a set of coordinates for each concept and each entity in a multi-dimensional Euclidean space; and means for displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the multi-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 11

A system for identifying and displaying inherent semantic constructs in entities in a data set with each entity comprising a web site, wherein the contents of the web pages that comprise said web site providing a description for said web site, with said constructs being held in common by more than one entity, in a manner such that the strength of semantic association between all entities and identified constructs are represented as visually-observable distances between data points in an n-dimensional Euclidean space, said system comprising: a network of computers, said network comprising interconnected server computers and client computers, some of said server computers storing and serving web pages having contents; means for producing a data set residing in said network, said data set comprising a plurality of data pairs, each data pair comprising a label for a web page and a textual or symbolic description of the web page; means for creating a set of modified web site descriptions by pruning irrelevant terms from web page descriptions and reducing each remaining term to a linguistic root form; means for extracting concepts from said modified web site descriptions residing in said network, each concept comprising a root term that is associated with more than one web site description; means for producing a similarity matrix residing in said network, wherein each web page is represented as a column and each concept is represented as a row or vise versa, with the element at the intersection of each such column and row containing a binary indication of whether the concept is found in the corresponding web page modified description; means for quantifying the associative structure of the data set by subjecting the similarity matrix to means for dual-scaling, in combination with auxiliary matrix transformation operations, thereby producing the coordinates of each concept and each web page in a multi-dimensional Euclidean space; and means for displaying the concepts as one type of virtual object and the web sites as a second type of virtual object with each object located at the appropriate coordinates in the multi-dimensional space, said means for displaying residing in a client computer, wherein the relative distances among the concept objects reflect the degree to which the concepts are associated with one another, wherein the relative distances among the entity objects reflect the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each web site object reflects the degree to which the content of each web page is associated with each concept.

Claim 12

A system for identifying and displaying inherent semantic constructs in entities in a data set with each entity comprising a web site, wherein the contents of the web pages that comprise said web site providing a description for said web site, with said constructs being held in common by more than one entity, in a manner such that the strength of semantic association between all entities and identified constructs are represented as visually-observable distances between data points in an n-dimensional Euclidean space, said system comprising: a network of computers, said network comprising interconnected server computers and client computers, some of said server computers storing and serving web pages having contents; means for producing a data set residing in said network, said data set comprising a plurality of data pairs, each data pair comprising a label for a web page and a textual or symbolic description of the web page; means for creating a set of modified web site descriptions by pruning irrelevant terms from web page descriptions and reducing each remaining term to a linguistic root form; means for extracting concepts from said modified web site descriptions residing in said network, each concept comprising a root term that is associated with more than one web site description; means for producing a similarity matrix residing in said network, wherein each web page is represented as a column and each concept is represented as a row or vise versa, with the element at the intersection of each such column and row containing a binary indication of whether the concept is found in the corresponding web page modified description; means for quantifying the associative structure of the data set by subjecting the similarity matrix to means for dual-scaling, in combination with auxiliary matrix transformation operations, thereby producing the coordinates of each concept and each web page in a multi-dimensional Euclidean space; and means for displaying the concepts as one type of virtual object and the web sites as a second type of virtual object with each object located at the appropriate coordinates in the multi-dimensional space, said means for displaying residing in a client computer, wherein the relative distances among the concept objects reflect the degree to which the concepts are associated with one another, wherein the relative distances among the entity objects reflect the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each web site object reflects the degree to which the content of each web page is associated with each concept.

Claim 31

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space; wherein said assembling step comprises producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; wherein said subjecting step and transforming step comprise quantifying the associative structure of the data set by subjecting the contingency matrix to procedures comprising dual-scaling, in combination with a matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space; and wherein said displaying step comprises displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 32

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space; wherein said assembling step comprises producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; wherein said subjecting step and transforming step comprise quantifying the associative structure of the data set by subjecting the contingency matrix to procedures comprising dual-scaling, in combination with a matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space; and wherein said displaying step comprises displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 36

A system for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: means for assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; means for normalizing the contingency matrix to produce a normalized matrix; means for subjecting the normalized matrix to singular value decomposition to produce singular values; means for transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; means for adjusting the range of said coordinates so that said space fits in a display frame; and means for displaying concepts and entities as objects in said space; wherein said means for assembling comprises means for producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; wherein said means for subjecting and means for transforming comprise means for quantifying the associative structure of the data set that is operative to subject the contingency matrix to means for dual-scaling, in combination with means for matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space; and wherein said means for displaying comprises means for displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 37

A system for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: means for assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; means for normalizing the contingency matrix to produce a normalized matrix; means for subjecting the normalized matrix to singular value decomposition to produce singular values; means for transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; means for adjusting the range of said coordinates so that said space fits in a display frame; and means for displaying concepts and entities as objects in said space; wherein said means for assembling comprises means for producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; wherein said means for subjecting and means for transforming comprise means for quantifying the associative structure of the data set that is operative to subject the contingency matrix to means for dual-scaling, in combination with means for matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space; and wherein said means for displaying comprises means for displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 38

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space; wherein said displaying step comprises displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 39

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space; wherein said displaying step comprises displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 40

A system for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: means for assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; means for normalizing the contingency matrix to produce a normalized matrix; means for subjecting the normalized matrix to singular value decomposition to produce singular values; means for transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; means for adjusting the range of said coordinates so that said space fits in a display frame; and means for displaying concepts and entities as objects in said space; wherein said means for displaying comprises means for displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 41

A system for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: means for assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; means for normalizing the contingency matrix to produce a normalized matrix; means for subjecting the normalized matrix to singular value decomposition to produce singular values; means for transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; means for adjusting the range of said coordinates so that said space fits in a display frame; and means for displaying concepts and entities as objects in said space; wherein said means for displaying comprises means for displaying the concepts as one type of virtual object and the entities as a second type of virtual object with each object located at the appropriate coordinates in the three-dimensional space, wherein the relative distance between each pair of concept objects reflects the degree to which the concepts are associated with one another, wherein the relative distance between each pair of entity objects reflects the degree to which the entities are associated with one another and wherein the relative distance between each concept object and each entity object reflects the degree to which each entity is associated with each concept.

Claim 42

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space; wherein said assembling step comprises producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; and wherein said subjecting step and transforming step comprise quantifying the associative structure of the data set by subjecting the contingency matrix to procedures comprising dual-scaling, in combination with a matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space.

Claim 43

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space; wherein said assembling step comprises producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; and wherein said subjecting step and transforming step comprise quantifying the associative structure of the data set by subjecting the contingency matrix to procedures comprising dual-scaling, in combination with a matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space.

Claim 44

A system for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: means for assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; means for normalizing the contingency matrix to produce a normalized matrix; means for subjecting the normalized matrix to singular value decomposition to produce singular values; means for transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; means for adjusting the range of said coordinates so that said space fits in a display frame; and means for displaying concepts and entities as objects in said space; wherein said means for assembling comprises means for producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; and wherein said means for subjecting and means for transforming comprise means for quantifying the associative structure of the data set that is operative to subject the contingency matrix to means for dual-scaling, in combination with means for matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space.

Claim 45

A system for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: means for assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; means for normalizing the contingency matrix to produce a normalized matrix; means for subjecting the normalized matrix to singular value decomposition to produce singular values; means for transforming the singular values to produce the coordinates for concept and entities in a simulated three-dimensional space; means for adjusting the range of said coordinates so that said space fits in a display frame; and means for displaying concepts and entities as objects in said space; wherein said means for assembling comprises means for producing a contingency matrix wherein each entity is represented as a column in said matrix and each concept is represented as a row, or vise versa, with the element at the intersection of each said column and each said row containing an indication of whether the corresponding concept is present in the corresponding modified entity description; and wherein said means for subjecting and means for transforming comprise means for quantifying the associative structure of the data set that is operative to subject the contingency matrix to means for dual-scaling, in combination with means for matrix transformation, thereby producing a set of coordinates for each concept and each entity in a three-dimensional Euclidean space.

Claim 52

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concepts and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space in which the degree to which the concepts and the items are related to one another is indicated to the user by Euclidian distances between each two objects.

Claim 53

A method for facilitating recognition and understanding of relationships or associations among a set of data entities comprising: assembling a contingency matrix that reflects the occurrence within descriptions of the entities of concepts identified in descriptions of the entities; normalizing the contingency matrix to produce a normalized matrix; subjecting the normalized matrix to singular value decomposition to produce singular values; transforming the singular values to produce the coordinates for concepts and entities in a simulated three-dimensional space; adjusting the range of said coordinates so that said space fits in a display frame; and displaying concepts and entities as objects in said space in which the degree to which the concepts and the items are related to one another is indicated to the user by Euclidian distances between each two objects.