Citation network viewer and method
View Patent ↗A visualization-based interactive legal research tool that generates from a multi-dimensional citation network a semantics-constrained citation sub-network that focuses on one individual issue in which a user is interested, and puts the sub-network on an interactive user interface (“UI”), which allows the researcher to browse, navigate, and jump over to start new sub-networks on different issues that are relevant to original issues.
1. A computer-implemented citation network viewer comprising:
data preparation means for taking a corpus of documents as input, and creating a series of metadata files, wherein each document has a citation and discusses at least one issue, and wherein relations between the documents as defined by their citations form a multi-dimensional citation network, the data preparation means including:
means for cutting each document into sentences;
means for identifying in each cut document, all citations, and all reasons-for-citing and headnotes associated with the identified citations, using a reason-for-citing algorithm;
means for converting the identified reasons-for-citing and the identified headnotes in each document into vectors; and
means for establishing semantic links between documents by pairing staffing reasons-for-citing in citing documents with cited reasons-for-citing and headnotes in cited documents by defining the text in each cited document that is closest semantically to the staffing reason-for-citing in the citing document, as defined by the equation:
RText
⇒
Max
K
i
=
1
(
Sim
(
V
i
,
SV
)
)
,
where K is the number of all reasons-for-citing and headnotes in each cited document, V is the vector for the i th reason-for-citing or headnote, and SV is the vector of the staffing reason-for-citing in the citing document; and
run-time means for carrying out run time operations, using the metadata files created by the data preparation means.
2. A computer-implemented citation network viewer comprising:
data preparation means for creating a series of metadata files from a corpus of input documents, each document having a citation and discussing at least one issue, wherein the citations of the documents form a multi-dimensional network of citations, and wherein the at least one issue is represented by one of a headnote and a reason for citing;
semantics-driven network traversing function means for creating from the metadata files of the documents forming the multi-dimensional network, a sub-network of citations of documents that focuses on a specific issue selected by a researcher; and
visualization-based interactive user interface means for displaying interactive graphics representing the sub-network;
wherein the data preparation means includes:
means for cutting each document into sentences;
means for identifying in each cut document, all citations, and all reasons-for-citing and headnotes associated with the identified citations, using a reason-for-citing algorithm;
means for converting the identified reasons-for-citing and the identified headnotes in each document into vectors; and
means for establishing semantic links between documents by pairing starting reasons-for-citing in citing documents with cited reasons-for-citing and headnotes in cited documents by defining the text in each cited document that is closest semantically to the starting reason-for-citing in the citing document, as defined by the equation:
RText
⇒
Max
K
i
=
1
(
Sim
(
V
i
,
SV
)
)
,
where K is the number of all reasons-for-citing and headnotes in each cited document, V is the vector for the i th reason-for-citing or headnote, and SV is the vector of the staffing reason-for-citing in the citing document.
3. The citation network viewer of claim 2 , wherein the sub-network comprises a plurality of nodes, each node representing one of a reason for citing or a headnote in a single document discussing the specific issue.
4. The citation network viewer of claim 2 , further comprising means for linking each citation to a vector produced from the reasons-for-citing or headnote associated with that citation.
5. The citation network viewer of claim 2 , further comprising means for building vector metadata files for reasons-for-citing and headnotes in each document, wherein each line in the vector metadata files contains information of one reason-for-citing or headnote represented by terms found in term lists and weights assigned to them.
6. The citation network viewer of claim 2 , wherein the run-time means includes
traversing function means for performing a depth-first search in the citation network represented by the semantic links between documents established by the means for establishing semantic links, and for retrieving forward-chained and backward-chained reasons-for-citing and headnotes based on a staffing reason-for-citing representing a user-specified citation, and
visualization-based interactive user interface means for displaying the retrieved reasons-for-citing and headnotes in the form of a sub-network of the citation network and for allowing the researcher to browse, navigate, and explore the contents of the sub-network.
7. The citation network viewer of claim 6 , wherein the depth-first search is set to allow for retrieval of reasons-for-citing and headnotes that are highly relevant to the staffing reason-for-citing along the citation chains, both directly and indirectly.
8. The citation network viewer of claim 6 , wherein the visualization-based interactive user interface also displays a digest window that allows the researcher to see other issues discussed in each document and “jump” to one of the other issues to staff a new sub-network.