Computing features of structured data

Patent metadata

Field	Value
Publication number	US-10896229-B2
Application number	US-201816188210-A
Country	US
Kind code	B2
Filing date	Nov 12, 2018
Priority date	Jul 8, 2014
Publication date	Jan 19, 2021
Grant date	Jan 19, 2021

Abstract

The present invention extends to methods, systems, and computer program products for computing features of structured data. Aspects of the invention include computing features of table components (e.g., of rows, columns, cells, etc.). Computed features can be used for ranking the table components. When aggregated, features for different components of a table can be used for ranking the table (e.g., a web table).

First claim

What is claimed is:

1. A computer-implemented method for computing features of structured data, comprising:
accessing, by a computing device, a table, the table including a subject column, a non-subject column, and a plurality of rows, cells of the table being at intersections between columns and rows of the table, the table annotated with additional content relevant to describing the table;
generating, by the computing device, an index for the table by indexing at least over data within the table along with the additional content;
storing, by the computing device, the index within a database, the index improving a relevance of providing the table in search results;
deriving, by the computing device, a semantic attribute for a first cell at an intersection of the non-subject column and a row of the table, wherein the semantic attribute comprises a first value in the first cell, a second value in a second cell at the intersection of the subject column and the row, and a name value for the non-subject column;
calculating, by the computing device, a first occurrence metric for the semantic attribute, the first occurrence metric indicating how frequently the semantic attribute is included in a plurality of other tables;
calculating, by the computing device, a second occurrence metric for the semantic attribute, the second occurrence metric indicating how frequently the semantic attribute is included in a web page that is selected from presented search results;
determining, by the computing device, a feature of the table by aggregating the semantic attribute, the first occurrence metric, and the second occurrence metric, wherein the feature of the table indicates a popularity or a trustworthiness of the table relative to the plurality of other tables;
creating, by the computing device, a ranking for the table based at least in part on the feature, wherein a more popular table is ranked higher than a less popular table, and wherein a more trustworthy table is ranked higher than a less trustworthy table;
receiving, by the computing device, a search query; and
surfacing, by the computing device, results from the table that satisfy the search query based at least in part on the ranking in order to distinguish the table from at least one other table that also satisfies the search query.

2. The computer-implemented method of claim 1 , wherein deriving the semantic attribute comprises combining the first value in the first cell, the second value in the second cell at the intersection of the subject column and the row, and the name value for the non-subject column into the semantic attribute.

3. The computer-implemented method of claim 1 , wherein deriving the semantic attribute comprises combining an entity-column name-value triple for the row.

4. The computer-implemented method of claim 3 , wherein:
deriving the semantic attribute comprises generating an entity-attribute binary table for the row and formulating the entity-column name-value triple from the entity-attribute binary table, the entity-attribute binary table having a first column corresponding to the subject column; and
the method further comprises counting an occurrence rate of the entity-column name-value triple across entity-attribute binary tables generated for one or more other tables.

5. The computer-implemented method of claim 4 , wherein the table comprises a second non-subject column, and wherein the method further comprises:
deriving an additional semantic attribute for a further cell at the intersection of the second non-subject column and the row, the additional semantic attribute providing additional information to distinguish the further value in the further cell from the further value in other cells;
calculating another frequency with which the additional semantic attribute is included in a second plurality of other tables; and
calculating another relevance of the other semantic attribute based on inclusion of the other semantic attribute in the second plurality of other tables and based on features of the second plurality of other tables.

6. The computer-implemented method of claim 5 , wherein:
the method further comprises calculating an additional frequency with which the other semantic attribute was selected from presented search results; and
determining the feature of the table comprises aggregating the additional frequency of the other semantic attribute into the feature.

7. The computer-implemented method of claim 1 , further comprising:
deriving an additional semantic attribute for a further cell at the intersection of the non-subject column and another row, the other semantic attribute providing additional information to distinguish the further value in the further cell from the further value in other cells;
calculating another frequency with which the other semantic attribute is included in a second plurality of other tables; and
calculating another relevance of the other semantic attribute based on inclusion of the other semantic attribute in the second plurality of other tables and based on features of the second plurality of other tables.

8. The computer-implemented method of claim 1 , wherein:
the method further comprises calculating an additional frequency with which the semantic attribute was selected from presented search results; and
determining the feature of the table comprises aggregating the additional frequency of the semantic attribute into the feature.

9. The computer-implemented method of claim 1 , wherein determining the feature of the table comprises determining at least one of a trustworthiness of the table or a popularity of the table.

10. A system for computing features of structured data, comprising:
one or more processors; and
memory comprising instructions that are executable by the one or more processors to perform operations comprising:
accessing, by a computing device, a table, the table including a subject column, a non-subject column, and a plurality of rows, cells of the table being at intersections between columns and rows of the table, the table annotated with additional content relevant to describing the table;
generating, by the computing device, an index for the table by indexing at least over data within the table along with the additional content;
storing, by the computing device, the index within a database, the index improving a relevance of providing the table in search results;
deriving, by the computing device, a semantic attribute for a first cell at an intersection of the non-subject column and a row of the table, wherein the semantic attribute comprises a first value in the first cell, a second value in a second cell at the intersection of the subject column and the row, and a name value for the non-subject column;
calculating, by the computing device, a first occurrence metric for the semantic attribute, the first occurrence metric indicating how frequently the semantic attribute is included in a plurality of other tables;
calculating, by the computing device, a second occurrence metric for the semantic attribute, the second occurrence metric indicating how frequently the semantic attribute is included in a web page that is selected from presented search results;
determining, by the computing device, a feature of the table by aggregating the semantic attribute, the first occurrence metric, and the second occurrence metric, wherein the feature of the table indicates a popularity or a trustworthiness of the table relative to the plurality of other tables;
creating, by the computing device, a ranking for the table based at least in part on the feature, wherein a more popular table is ranked higher than a less popular table, and wherein a more trustworthy table is rank

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Zoryn Kanstantsyn
• Chen Zhimin
• Chakrabarti Kaushik
• Finnigan James P
• Narasayya Vivek R
• Chaudhuri Surajit
• Ganjam Kris

Classifications

• G06F16/951 Primary

  Indexing; Web crawling techniques · CPC title

• G06F16/2458

  Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries · CPC title