Attribute extraction
View Patent ↗A system for extracting attributes can analyze text from data sources, extract n-grams from the text as candidate attribute and service/product pairs, prompt a human operator to rate the suitability of the candidate attribute and service/product pairs, and, based on the ratings, add the candidate attribute and service/product pairs to an attribute dictionary. In embodiments, an attribute extraction system includes an n-gram extractor, an attribute selector interface, and a dictionary builder. Data sources may include product titles, category descriptions, product descriptions, and like data from one or more product databases. In embodiments, the attribute dictionary is analyzed to determine canonical names for products or services and name variants for the products or services.
1. A computer-implemented method for extracting attributes, comprising:
receiving a text passage from a data source;
at an n-gram extractor, identifying n-grams in the text passage;
at the n-gram extractor, selecting a candidate attribute and service/product pair from the n-grams in the text passage;
at an attribute selector interface, transmitting the candidate attribute and service/product pair to a human operator;
at the attribute selector interface, prompting the human operator to input an attribute suitability rating for the candidate attribute and service/product pair;
at the attribute selector interface, receiving an input entered by the human operator, the input corresponding to the attribute suitability rating;
at a dictionary builder, receiving the attribute suitability rating for the candidate attribute and service/product pair; and
at the dictionary builder, if the attribute suitability rating exceeds a suitability threshold, adding the candidate attribute and service/product pair to an attribute dictionary.
2. The method of claim 1 , wherein the data source comprises one selected from a group consisting of a product title, a category description, and a product description.
3. The method of claim 1 , wherein the attribute suitability rating comprises a numerical score.
4. The method of claim 1 , wherein the attribute suitability rating comprises a dichotomous rating.
5. The method of claim 1 , further comprising analyzing the attribute dictionary to determine a canonical name for a product or a service of the candidate attribute and service/product pair.
6. The method of claim 1 , further comprising analyzing a context portion of the text passage to determine a name variation of a product or a service of the candidate attribute and service/product pair.
7. The method of claim 1 , wherein the candidate attribute and service/product pair comprises a high frequency n-gram.
8. The method of claim 1 , further comprising at the dictionary builder, generating an ontological graph having nodes that represent products or services.
9. A computer system for extracting attributes comprising:
an attribute extractor comprising a memory and a processing device, the memory storing computer-readable instructions directing the processing device to:
receive a text passage from a data source;
identify n-grams in the text passage;
select a candidate attribute and service/product pair from the n-grams in the text passage;
transmit the candidate attribute and service/product pair to a human operator interface;
receive an input entered by the human operator, the input corresponding to an attribute suitability rating for the candidate attribute and service/product pair; and
if the attribute suitability rating exceeds a suitability threshold, add the candidate attribute and service/product pair to an attribute dictionary.
10. The computer system of claim 9 , wherein the data source comprises one selected from a group consisting of a product title, a category description, and a product description.
11. The computer system of claim 9 , wherein the attribute suitability rating comprises a numerical score.
12. The computer system of claim 9 , wherein the attribute suitability rating comprises a dichotomous rating.
13. The computer system of claim 9 , wherein the computer-readable instructions further direct the processing device to analyze the attribute dictionary to determine a canonical name for a product or a service of the candidate attribute and service/product pair.
14. The computer system of claim 9 , wherein the computer-readable instructions further direct the processing device to analyze a context portion of the text passage to determine a name variation of a product or a service of the candidate attribute and service/product pair.
15. A computer system for extracting attributes comprising:
an n-gram extractor comprising an n-gram extractor memory and an n-gram extractor processing device, the n-gram extractor memory storing computer-readable instructions directing the n-gram extractor processing device to extract a candidate attribute and service/product pair from a text passage;
an attribute selector interface comprising an attribute selector interface memory and an attribute selector interface processing device, the attribute selector interface memory storing computer-readable instructions directing the attribute selector interface processing device to:
present an interface to a human operator and
prompt the human operator to input an attribute suitability rating of the candidate attribute and service/product pair;
a dictionary builder comprising a dictionary builder memory and a dictionary builder processing device, the dictionary builder memory storing computer-readable instructions directing the dictionary builder processing device to:
receive the attribute suitability rating input by the human operator;
based on the attribute suitability rating, determine if the candidate attribute and service/product pair comprises an attribute and service/product pair; and
if the candidate attribute and service/product pair comprises an attribute and service/product pair, add the candidate attribute and service/product pair to an attribute dictionary.
16. The computer system of claim 15 , wherein the n-gram extractor memory stores computer-readable instructions further directing the n-gram extractor processing device to retrieve the text passage from a data source.
17. The computer system of claim 15 , wherein if the candidate attribute and service/product pair comprises the attribute and service/product pair, the instructions further direct the dictionary builder to store metadata relating to the attribute and service/product pair, and
wherein the metadata comprises:
a frequency of the attribute and service/product pair in a data source; and
a ranking of each attribute and service/product pair from the human operator.
18. The computer system of claim 15 , wherein the text passage is extracted from a data source comprising one selected from a group consisting of a product title, a category description, and a product description.
19. The computer system of claim 15 , wherein the attribute suitability rating comprises a dichotomous rating.
20. The computer system of claim 15 , further comprising an analyzer adapted to analyze the attribute dictionary to determine a canonical name for a product or a service of the attribute and service/product pair.