CATEGORY-SENSITIVE RANKING FOR TEXT

• US 20120278332A1
• Filed: 12/30/2009
• Published: 11/01/2012
• Est. Priority Date: 12/30/2009
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Wang, Yi, Tao, Bo, Liu, Zhiyuan
• Granted Patent: US 9,092,422 B2
• Time in Patent Office: Days
• US Class Current: 707/740
• CPC Class Codes: G06F 16/345 Summarisation for human users
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates

Abstract

Provided are methods, systems and apparatus which include computer program products, for generating topic models for text summarization In one aspect, a method includes receiving a first document of text that is associated with one or more category labels and that includes one or more sequences of one or more words, determining a category label that represents a first category associated with the first document, sampling the one or more sequences to determine a topic and a co-occurrence relationship between the topic and the category label, where a topic represents a subdivision within a category, sampling the one or more sequences to determine a co-occurrence relationship between a sequence in the first document and the topic, and generating a category-topic model that represents the co-occurrence relationships.

Claims

1. A method comprising:

receiving a first document of text that is associated with one or more category labels and that includes one or more sequences of one or more words;

determining a category label that represents a first category associated with the first document;

sampling the one or more sequences to determine a topic and a co-occurrence relationship between the topic and the category label, where a topic represents a subdivision within a category;

sampling the one or more sequences to determine a co-occurrence relationship between a sequence in the first document and the topic; and

generating a category-topic model that represents the co-occurrence relationships.

2. The method of claim 1, where the generating includes:

determining, from the co-occurrence relationship between the topic and the category label, a first probability of the topic being associated with a document given that the category label is associated with the document; and

determining, from a co-occurrence relationship between the sequence and the topic, a second probability of the sequence occurring in the document given that the topic is associated with the document.

3. The method of claim 2, where determining the co-occurrence relationships includes:

calculating a frequency of co-occurrence of the topic and the category label in the document; and

calculating a frequency of co-occurrence of the sequence and the topic in the document.

4. The method of claim 3, where the sampling is represented using the expression:

5. The method of claim 4, where each sampling is performed independently on multiple processors, each processor p calculates an update Δ

C_|p^TC and an update Δ

C_|p^TC, and the frequencies of co-occurrences are updated according to the equations

6. The method of claim 4, where the determining the first and second probabilities includes:

calculating the conditional probabilities that the topic is k given that the category label is c, the sequence is v given that the topic is k, the category label is c given that the topic is k, and the topic is k given that the sequence is v.

7. The method of claim 6, where the conditional probabilities satisfy the expressions:

8. The method of claim 1, where the category-topic model represents co-occurrence relationships between topics and category labels determined by sampling one or more sequences of one or more words in each of a collection of received documents, and co-occurrence relationships between the sequences in each document and the topics determined by sampling the one or more sequences of one or more words in each of the collection of received documents.

9. The method of claim 1, further comprising:

receiving a textual input;

determining in the textual input one or more sequences of one or more words that are each associated with a respective category;

ranking each of the sequences using the category-topic model;

identifying one or more of the determined sequences that have a ranking greater than a threshold ranking value; and

generating a summary of the textual input that includes the identified sequences.

10. The method of claim 9, where the ranking includes calculating a probabilistic characteristic value for each sequence of one or more words, where the probabilistic characteristic value is expressed as:

11. The method of claim 1, where the category labels are human-readable.

12. The method of claim 1, where at least one of the topics is not human-readable.

13. The method of claim 1, where a sequence of one or more words is a phrase, sentence, or paragraph.

14. The method of claim 1, where the first document includes category labels that represent the one or more categories, and determining the category label includes selecting one of the category labels.

15. The method of claim 1, where determining the category label includes assigning a unique label to the first document.

16. A method comprising:

receiving a textual input;

receiving a category-topic model that represents co-occurrence relationships between topics and category labels and co-occurrence relationships between sequences of one or more words and topics;

determining in the textual input one or more sequences of one or more words that are each associated with a respective category label;

ranking each of the determined sequences using the category-topic model;

identifying one or more of the sequences that have a ranking greater than a threshold ranking value; and

generating a summary of the textual input that includes the identified sequences.

17. The method of claim 16, where the category-topic model is produced by a method in accordance with methods 1 through 8.

18. A system comprising:

a machine-readable storage device including a program product; and

one or more processors operable to execute the program product and perform operations comprising;

receiving a textual input;

receiving a category-topic model that represents co-occurrence relationships between topics and category labels and co-occurrence relationships between sequences of one or more words and topics;

determining in the textual input one or more sequences of one or more words that are each associated with a respective category label;

ranking each of the determined sequences using the category-topic model;

identifying one or more of the sequences that have a ranking greater than a threshold ranking value; and

generating a summary of the textual input that includes the identified sequences.