Extracting similar entities from lists / tables
AssigneeMicrosoft Corporation
InventorsSreenivas Gollapudi, Alan Halverson
Life of this patent
Estimating the term…
Abstract
Large numbers of lists of entities may be mined for similar entities to related searches. A representation for each list may be determined to provide for a comparison between lists and to support membership checks. A score for an element in a list may be computed that represents the validity of an item in the corpus of lists. Thus, a spurious element would receive a very low score, where a valid element would receive a higher score. A list weight is then computed using the constituent element weights, and the element and list weight are used to compute the nearest neighbors of a given query element.
Claims20
A method for determining similar entities to a query element from a corpus of lists, comprising:
- determining a representation for each list in the corpus;
- determining a score for each element on each list;
- determining a weight for each element and each list;
- and
- determining nearest neighbors to the query element in accordance with the weight of each element and the weight of each list.
A system of determining nearest neighbors of a query element, comprising:
- a list gathering engine to maintain a corpus of lists;
- a comparison engine that determines a representation of each list and compares the lists to each other;
- a weighting engine that determines weights of list elements and lists;
- and
- a ranking engine that determined the nearest neighbors from the lists.
A method for determining similar entities in HTML tables, comprising:
- determining a sketch of the HTML tables;
- evaluating each row as a list of column values;
- and
- characterizing each HTML table as a list of sketches.
Dependencies are inferred from the claim text.