Detection and creation of appropriate row concept during automated model generation

Abstract

A system and method for assigning concepts to sets of values. Data is received, wherein the data is organized as a plurality of named fields and as two or more sets of values associated with the named fields, wherein each named field is assigned to a category. For each category, determine whether there is at least one identifier field for that category, wherein each identifier field is a named field that acts as an identifier for that category, and identify identifier fields, if any, for each category that have a unique value in the identifier field for that category for each set of values. Then select one of the categories as a concept representing the sets of values. In some embodiments, the data is organized as a table, wherein the named fields are columns and the sets of values are rows.

Description

technical field [0001] The present invention relates to natural language analysis, and more particularly to classification of data in data sets. Background technique [0002] Accurately converting tabular data into useful query models can be difficult. Often, specialized modeling of the data is required, and the analysis tools required for conversion often require training and specialization that are not common among business users. [0003] Of course, there are many challenges in creating such a tool. If automated modeling does not reflect the data or knowledge it represents, the queries it may generate may not be useful in answering users' questions. If users' questions cannot be parsed and understood by the system, the system cannot accurately generate queries to answer their questions. Over the past 50 years, accurate natural language parsing has become a branch of computer science that is still considered in its infancy. [0004] In traditional analysis systems, the...

AI Technical Summary

Problems solved by technology

But this is done by someone who understands the data they are modeling, and takes time

Images