Intelligent autofill

• US 7,254,569 B2
• Filed: 06/23/2004
• Issued: 08/07/2007
• Est. Priority Date: 05/12/2004
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Goodman, Joshua T, Kadie, Carl M, Chickering, David M, Bradford, Donald E, Glasgow, Dane A
• Primary Examiner(s): Rimell, Sam
• Assistant Examiner(s): Lewis, Alicia M
• Application Number:
  US10/874,955
• Publication Number:
  US 20050257148A1
• Time in Patent Office: 1,140 Days
• Field of Search: 707/1, 707/3, 707/104.1, 715/780, 715/507
• US Class Current: 1/1
• CPC Class Codes: G06F 40/174 Form filling; Merging
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

The present invention provides a unique system and method that can employ machine learning techniques to automatically fill one or more fields across a diverse array of web forms. In particular, one or more instrumented tools can collect input or entries of form fields. Machine learning can be used to learn what data corresponds to which fields or types of fields. The input can be sent to a central repository where other databases can be aggregated as well. This input can be provided to a machine learning system to learn how to predict the desired outputs. Alternatively or in addition, learning can be performed in part by observing entries and then adapting the autofill component accordingly. Furthermore, a number of features of database fields as well as constraints can be employed to facilitate assignments of database entries to form values—particularly when the web form has never been seen before by the autofill system.

Claims

1. Computer-executable instructions for performing a method that facilitates automatic data entry of forms, the computer-executable instructions stored on one or more computer readable media, the method comprising:

receiving a web-based form having a plurality of input fields;

determining a plurality of feature values that correspond to at least some of the plurality of input fields;

using a machine learned model to generate two or more assignments of field types to the at least some of the plurality of input fields;

determining a joint probability of each of the two or more assignments consistent with one or more constraints; and

automatically filling in at least one of the plurality of input fields with input data from a database according to the one of the two or more assignments having the highest joint probability.

2. The computer-executable instructions of claim 1, at least some of the plurality of input fields comprising a text field, file input field, select box, radio button, or check box or combinations thereof.

3. The computer-executable instructions of claim 1, the machine learned model learned by a method comprising:

observing data input by a user relating to one or more known input fields to learn what input data corresponds to which of the one or more known input fields.

4. The computer-executable instructions of claim 1, the one or more constraints comprising reuse constraints that limit a number of times any one database entry is used per web-based form.

5. The computer-executable instructions of claim 1, the one or more constraints comprising composition constraints that control which input data are allowed to be present concurrently on any web-based form.

6. The computer-executable instructions of claim 1, the one or more constraints comprising constraints on input fields occurring together or in a particular order.

7. The computer-executable instructions of claim 1, at least one feature value corresponding to a particular input field being above, below, to the left of, to the right of, immediately below, immediately above, immediately to the left of, or immediately to the right of some other input field or combinations thereof.

8. The computer-executable instructions of claim 3, further comprising using one or more instrumented toolbars to observe the data input the by user.

9. The computer-executable instructions of claim 1, at least some of the plurality of feature values relating to:

MAXLENGTH or SIZE entries;

substrings of a field name;

information from one or more adjacent form fields;

nearby text or substrings of nearby text or combinations thereof.

10. The computer-executable instructions of claim 1, the plurality of feature values corresponding to Boolean values or numeric values.

11. The computer-executable instructions of claim 10, wherein determining the plurality of features values that correspond to at least some of the plurality of input fields comprises:

finding useful ranges; and

looking for common and useful character sequences.

12. The computer-executable instructions of claim 1, wherein automatically filling in at least one of the plurality of input fields with input data from a database comprises providing autofill suggestions:

in a drop down box from a field when a user begins to enter data into the field;

or

as soon as a user enters the field, filling the field with suggested data or both.

13. The computer-executable instructions of claim 1, further comprising detecting whether the web-based form comprises multiple varieties of input fields and separating them into at least two subsets of fields.

14. The computer-executable instructions of claim 13, wherein detecting whether the web-based form comprises multiple varieties of input fields and separating them into at least two subsets of fields comprises employing Heuristics to detect the multiple sets and to separate them into at least two subsets.

15. The computer-executable instructions of claim 1, further comprising observing user entries of input fields to further learn the machine learned model.

16. The computer-executable instructions of claim 15, further comprising not learning user entries for input fields that appear to comprise sensitive information.

17. The computer-executable instructions of claim 15, the machine learned model trained using an offline algorithm that includes collecting a plurality of observed instances of input field entries and storing them.

18. The computer-executable instructions of claim 15, the machine learned model trained using an online algorithm using any one of Naï

ves Bayes approach, gradient descent, winnow, or perceptron algorithm.