System and method for correction of acquired transaction text fields
AssigneeIntuit Inc.
InventorsYair Horesh, Yehezkel S. Resheff, Rotem Rozenblum, Shimon Shahar
Life of this patent
Estimating the term…
Abstract
Transactions include text fields, such as description fields. Transactions are extracted from financial institutions using web-scraping extraction. In the process of extracting transactions, errors can be introduced into text fields, such as the inclusion of a space within a word or the removal of a space between words. A statistical approach is applied to the text fields. When two alternative text fields are presented, the alternative that statistically includes more common tokens, such as unigrams and bigrams, is chosen as the correct alternative. The incorrect alternative is replaced by the correct alternative in the text field.
Claims18
A computing system implemented method comprising:
- generating first stripped text field data from text field data based on removing numeric characters from the text field data;
- determining, after removing the numeric characters, unigram count data for each unigram within the first stripped text field data;
- determining, after removing the numeric characters, bigram count data for each bigram within the first stripped text field data;
- generating second stripped text field data from the text field data based on removing numeric characters and delimiter characters from the text field data, the delimiter characters including at least one of a tab, a comma, or a semicolon;
- detecting an inconsistency within the second stripped text field data;
- determining a pair ratio for equivalent unigrams and bigrams within the first stripped text field data based on the unigram count data and the bigram count data;
- and
- generating, in response to detecting the inconsistency, corrected text field data based on the pair ratio.
A system comprising:
- at least one processor;
- and
- at least one memory coupled to the at least one processor, the at least one memory storing instructions that, when executed by the at least one processor, cause the system to:
- generate first stripped text field data from text field data based on removing numeric characters from the text field data;
- determine, after removing the numeric characters, unigram count data for each unigram within the first stripped text field data;
- determine, after removing the numeric characters, bigram count data for each bigram within the first stripped text field data;
- generate second stripped text field data from the text field data based on removing numeric characters and delimiter characters from the text field data, the delimiter characters including at least one of a tab, a comma, or a semicolon;
- detect an inconsistency within the second stripped text field data;
- determine a pair ratio for equivalent unigrams and bigrams within the first stripped text field data based on the unigram count data and the bigram count data;
- and
- generate, in response to detecting the inconsistency, corrected text field data based on the unigram count data and the bigram count data.
A computing system implemented method comprising:
- generating first stripped text field data from text field data based on removing numeric characters from text field data;
- determining, after removing the numeric characters, unigrams and bigrams in the first stripped text field data;
- generating second stripped text field data from the text field data based on removing numeric characters and delimiter characters from the text field data, the delimiter characters including at least one of a tab, a comma, or a semicolon;
- detecting an inconsistency within the second stripped text field data;
- determining unigram count data and bigram count data for each unigram and bigram, respectively, within the first stripped text field data;
- determining a pair ratio for equivalent unigrams and bigrams within the first stripped text field data based on the unigram count data and the bigram count data;
- and
- generating, in response to detecting the inconsistency, corrected text field data based on the pair ratio.
Dependencies are inferred from the claim text.