Method and system for matching data sets of non-standard formats

• US 8,090,725 B1
• Filed: 04/16/2010
• Issued: 01/03/2012
• Est. Priority Date: 01/13/2006
• Status: Active Grant

• Current Assignee: CareerBuilder LLC (Nexstar Media Group, Inc.)
• Original Assignee: CareerBuilder LLC (Nexstar Media Group, Inc.)
• Inventors: Cranfill, Andrew B.
• Primary Examiner(s): Robinson, Greta
• Assistant Examiner(s): Weinrich, Brian E
• Application Number:
  US12/761,569
• Time in Patent Office: 627 Days
• Field of Search: 707/750, 707/804, 707/947
• US Class Current: 707/750
• CPC Class Codes: G06F 16/258 Data format conversion from or to a database
• CPC Class Codes: Y10S 707/947 Human resources

Abstract

A system and method is described for receiving a plurality of non-standardized data sets and generating respective plurality of standardized profiles that can be used for efficiently comparing and matching one profile against the other plurality of profiles. One application of this invention is to convert job seekers'"'"' resumes and job postings into respective profiles and then permitting either a job seeker to search for job postings that most closely match the job seeker'"'"'s resume or, conversely, permitting an employer to search for job seekers whose resumes most closely match the employer'"'"'s job posting.

Claims

1. A method for comparing a plurality of resumes, including the following steps:

receiving a first resume from a database stored on a computer;

parsing the first resume into bands based on a predefined setting;

generating a word array by parsing the text in each band into separate parsed words and storing each of the parsed words in the word array, wherein the word array includes separate rows for each of the parsed words and a column populated with information that is indicative of the band associated with each of the parsed words;

standardizing each of the words contained in the word array, by iteratively correcting punctuation, replacing well-known abbreviations, or removing common words;

generating an attribute array by iteratively comparing each of the parsed words contained in the word array to attributes in an attribute dictionary and adding each of the parsed words that match one of the attributes in the attribute dictionary to the attribute array, wherein the attribute array includes information regarding the number of times each of the attributes occurs within the first resume and information indicative of the band in which each of the attributes was first found;

identifying root attributes based on the number of times in which attributes or multi-word attributes occur within the first resume and counting the number of occurrences of each root attribute in the first resume;

identifying leaf attributes that are related to root attributes and counting the number of occurrences of each leaf attribute in the first resume;

generating a first metric indicative of the significance of each of the attributes in the attribute array to the first resume;

generating a second metric indicative of the significance of each of the root attributes to the first resume;

generating a third metric indicative of the significance of each of the leaf attributes to the associated root attribute;

weighting the first, second and third metrics, wherein for each attribute, the first, second and third metrics are one of the bands indicating a relative position of the attribute within the first resume, a number of occurrences of each of the associated root attributes in the first resume, and a support value indicating a relationship to each of the associated root attributes or a combination of the three metrics;

ranking the attributes based on a weighted value of the first metric for each of the attributes;

ranking the root attributes based on a weighted value of the second metric for each of the root attributes;

ranking the leaf attributes based on a weighted value of the third metric for each of the leaf attributes;

generating a profile for the first resume based on the rank of the attributes, the root attributes and the leaf attributes;

selecting one or more additional resumes for comparison and generating profiles for the additional resumes by using the same steps that were employed to generate the profile for the first resume;

comparing the profile for the first resume with the profiles for the additional resumes; and

ranking the profiles for the additional resumes based on how closely the profiles for the additional resumes match the profile for the first resume.

2. The method for comparing a plurality of resumes described in claim 1, wherein the predefined setting is based on a date reference, subject matter or location of text that populates the resume.

3. The method for comparing a plurality of resumes described in claim 1, wherein the step of parsing the text further includes the step of parsing the text in each band into words based on the occurrence of spaces, line feeds or carriage returns.

4. The method for comparing a plurality of resumes described in claim 1, wherein the step of generating the attribute array includes ignoring insignificant dependent concepts.

5. The method for comparing a plurality of resumes described in claim 1, wherein the step of generating the attribute array includes standardizing each of the words contained in the word array by separating each of the words in the word array into dependent and independent words.

6. The method for comparing a plurality of resumes described in claim 1, wherein the step of generating the attribute array includes identifying synonyms for each of the words in the word array.

7. The method for comparing a plurality of resumes described in claim 1, including the step of generating a multi-word attribute array by storing one or more groups of words in a multi-word buffer.

8. The method for comparing a plurality of resumes described in claim 7, wherein the multi-word buffer includes four words or less.

9. The method of comparing a plurality of resumes described in claim 8, wherein the step of generating a multi-word attribute array further includes iteratively comparing each of the groups of words to multi-word attributes in a multi-word attribute dictionary and adding groups of words that match one of the multi-word attributes to the multi-word attribute array.

10. The method of comparing a plurality of resumes described in claim 9, wherein the multi-word attribute array includes information regarding the number of times each of the groups of words in the multi-word attribute array occurs within the first resume and information indicative of the band in which each of the groups of words was first found.

11. The method for comparing a plurality of resumes described in claim 10, wherein the step of generating the multi-word attributes includes removing spam terms from the multi-word buffer.