Identifying information related to a particular entity from electronic sources, using dimensional reduction and quantum clustering
Presented are systems and methods for identifying information about a particular entity including acquiring electronic documents having unstructured text, that are selected based on one or more search terms from a plurality of terms related to the particular entity. Tokenizing the acquired documents to form a data matrix and then calculating a plurality of eigenvectors, using the data matrix and the transpose of the data matrix. The variance is then acquired for determining the amount of intra-clustering between the documents and then the acquired documents are clustered using some of the eigenvectors and the variance.
1. A system comprising:
one or more servers, each server having a processor and a memory, the one or more servers comprising:
a collector module configured to acquire a plurality of eigenvectors, each having a corresponding eigenvalue, wherein
the plurality of eigenvectors are based on a plurality of tokenized electronic documents having unstructured text, the plurality of tokenized electronic documents forming a data matrix, and
the unstructured text includes background terms and nonbackground terms; and
a dimensional reduction module configured to:
classify the plurality of eigenvectors and their corresponding eigenvalues into one or more background eigenvectors and background eigenvalues, and one or more nonbackground eigenvectors and nonbackground eigenvalues, wherein the background eigenvectors correspond to the background terms and the nonbackground eigenvectors correspond to nonbackground terms,
acquire a threshold,
compare the nonbackground eigenvalues with the threshold, and
provide the nonbackground eigenvectors whose corresponding nonbackground eigenvalues exceed the threshold, wherein the provided nonbackground eigenvectors are used for clustering the plurality of documents.
2. The system of claim 1 , wherein the dimensional reduction module is further configured to set the threshold by a user, by a human operator other than a user, or from a data storage.
3. The system of claim 1 , wherein the one or more servers further comprise a data processing module that is configured to:
construct a guess eigenvector;
multiply the guess eigenvector by the transpose of the data matrix to obtain a first vector, wherein the data matrix corresponds to the plurality of tokenized electronic documents having unstructured text;
multiply the first vector with the data matrix to obtain a second vector; and
provide the second vector if the second vector is an eigenvector.
4. The system of claim 1 , wherein the one or more servers further comprise a data processing module that is configured to:
construct a guess eigenvector to a distributed matrix, wherein the distributed matrix is related to the data matrix and the transpose of the data matrix;
partition the guess eigenvector into one or more sub-vectors;
distribute the one or more sub-vectors to one or more server slaves that contain at least one tokenized electronic document;
acquire from the one or more server slaves, a second vector that corresponds to the multiplication of the sub-vector, that was distributed to the server slave, with the tokenized electronic document contained on that server slave;
assemble the acquired vectors into a single reconstituted vector; and
provide the reconstituted vector if it is an eigenvector of the distributed matrix.
5. A method comprising:
acquiring at one or more servers, each of the servers having a processor and a memory, a plurality of eigenvectors, each having a corresponding eigenvalue, wherein
the plurality of eigenvectors are based on a plurality of tokenized electronic documents having unstructured text, the plurality of tokenized electronic documents forming a data matrix, and
the unstructured text includes background terms and nonbackground terms; and
classifying the plurality of eigenvectors and their corresponding eigenvalues into one or more background eigenvectors and background eigenvalues, and one or more nonbackground eigenvectors and nonbackground eigenvalues, wherein the background eigenvectors correspond to the background terms and the nonbackground eigenvectors correspond to nonbackground terms;
acquiring a threshold;
comparing the nonbackground eigenvalues with the threshold; and
providing the nonbackground eigenvectors whose corresponding nonbackground eigenvalues exceed the threshold, wherein the provided nonbackground eigenvectors are used for clustering the plurality of documents.
6. The method of claim 5 , wherein the threshold is set by a user, by a human operator other than a user, or from a data storage.
7. The method of claim 5 , wherein at least one of the acquired plurality of eigenvectors is generated by:
constructing a guess eigenvector;
multiplying the guess eigenvector by the transpose of the data matrix to obtain a first vector, wherein the data matrix corresponds to the plurality of tokenized electronic documents having unstructured text;
multiplying the first vector with the data matrix to obtain a second vector; and
providing the second vector if the second vector is an eigenvector.
8. The method of claim 5 , wherein at least one of the acquired plurality of eigenvectors is generated by:
constructing a guess eigenvector to a distributed matrix, wherein the distributed matrix is related to the data matrix and the transpose of the data matrix;
partitioning the guess eigenvector into one or more sub-vectors;
distributing the one or more sub-vectors to one or more server slaves that contain at least one tokenized electronic document;
acquiring from the one or more server slaves, a second vector that corresponds to the multiplication of the sub-vector that was distributed to the server slave with the tokenized electronic document contained on that server slave;
assembling the acquired vectors into a single reconstituted vector;
determining whether the reconstituted vector is an eigenvector of the distributed matrix; and
providing the reconstituted vector if it is an eigenvector of the distributed matrix.
9. A non-transitory computer-readable medium comprising program code, the program code being operable, when executed by a computer system, to cause the computer system to perform a method comprising:
acquiring a plurality of eigenvectors, each having a corresponding eigenvalue, wherein
the plurality of eigenvectors are based on a plurality of tokenized electronic documents having unstructured text, the plurality of tokenized electronic documents forming a data matrix, and
the unstructured text includes background terms and nonbackground terms; and
classifying the plurality of eigenvectors and their corresponding eigenvalues into one or more background eigenvectors and background eigenvalues, and one or more nonbackground eigenvectors and nonbackground eigenvalues, wherein the background eigenvectors correspond to the background terms and the nonbackground eigenvectors correspond to nonbackground terms;
acquiring a threshold;
comparing the nonbackground eigenvalues with the threshold; and
providing the nonbackground eigenvectors whose corresponding nonbackground eigenvalues exceed the threshold, wherein the provided nonbackground eigenvectors are used for clustering the plurality of documents.
10. The computer readable medium of claim 9 , wherein the threshold is set by a user, by a human operator other than a user, or from a data storage.
11. The computer readable medium of claim 9 , wherein at least one of the acquired plurality of eigenvectors is generated by:
constructing a guess eigenvector;
multiplying the guess eigenvector by the transpose of the data matrix to obtain a first vector, wherein the data matrix corresponds to the plurality of tokenized electronic documents having unstructured text;
multiplying the first vector with the data matrix to obtain a second vector; and
providing the second vector if the second vector is an eigenvector.
12. The computer readable medium of claim 9 , wherein at least one of the acquired plurality of eigenvectors is generated by:
constructing a guess eigenvector to a distributed matrix, wherein the distributed matrix is related to the data matrix and the transpose of the data matrix;
partitioning the guess eigenvector into one or more sub-vectors;
distributing the one or more sub-vectors to one or more server slaves that contain at least one tokenized electronic document;
acquiring from the one or more server slaves, a second vector that corresponds to the multiplication of the sub-vector that was distributed to the server slave with the tokenized electronic document contained on that server slave;
assembling the acquired vectors into a single reconstituted vector;
determining whether the reconstituted vector is an eigenvector of the distributed matrix; and
providing the reconstituted vector if it is an eigenvector of the distributed matrix.