Categorization of forms to aid in form search

1. In a computer network environment, a method for facilitating searching for a form by associating forms with categories based on form features, the method comprising:

receiving, by a processor of a computing device, a quantity of categories, each category indicating a form feature, wherein the form features include (i) non-text field characteristics indicating a relation between multiple field locations, and (ii) field-specific text characteristics indicating an association between a form field and a label of the form field;

determining a category feature vector for each of the categories, each respective category feature vector indicating form features associated with the respective category;

determining a form feature vector for a respective form from a set of forms, each respective form feature vector indicating form features associated with the respective form;

automatically associating, by the processor, the set of forms with respective categories of the quantity of categories, the association based on a similarity between the category feature vectors of the respective categories and the form feature vectors of the set of forms;

automatically determining, based on form features of a particular form of the set of forms, a combination of at least two categories to which the particular form belongs;

automatically adjusting the quantity of categories to include a combined category based on the combination of the at least two categories;

storing the forms and the respective categories associated with the forms at an electronic form search server; and

providing search results from the electronic form search server based on input identifying the combined category, a form identified as a search result based on the form being associated with the at least two categories.

2. The method of claim 1 , wherein associating forms with respective categories comprises:

creating a training model using a collection of forms; and

applying the training model to additional forms to associate the additional forms with the respective categories based on form features for the non-text field characteristics or field-specific text characteristics of the additional forms.

3. The method of claim 1 , wherein associating forms with respective categories comprises:

using a machine learning algorithm to create a training model using a collection of pre-categorized forms, the training model representing a significance of each of the form features to each of the categories; and

applying the training model to additional forms to associate the additional forms with the respective categories based on form features for the non-text field characteristics or field-specific text characteristics of the additional forms.

4. The method of claim 1 , wherein associating forms with respective categories comprises:

receiving user input categorizing forms in a collection;

creating a training model using the collection of categorized forms; and

applying the training model to uncategorized forms to categorize the uncategorized forms in the respective categories based on the form features for the non-text field characteristics or field-specific text characteristics of the forms.

5. The method of claim 1 , wherein associating forms with respective categories comprising determining a feature vector for individual forms, wherein respective feature vectors comprise a plurality of form features, wherein associating the forms with the categories is based on the feature vectors.

6. The method of claim 1 , wherein associating forms with respective categories is unsupervised and is based on distances between the form features of the forms.

7. The method of claim 1 , wherein storing the forms and the respective categories associated with the forms comprises at least one of: storing an indication of significance of a category to a form, or tagging each form with a best matching category.

8. The method of claim 1 , wherein:

determining the form feature vector includes identifying a field for each form, wherein the field is identified based on graphical characteristics of the set of forms or metadata associated with the set of forms, and

the non-text field characteristics and the field-specific text characteristics are based on the identified fields.

9. In a computer network environment, a method for facilitating searching for a form by associating forms with categories using a training model, the method comprising:

creating a training model using form features for non-text field characteristics or field-specific text characteristics of forms in a collection of forms, wherein the non-text field characteristics indicate a relation between multiple field locations, and the field-specific text characteristics indicate an association between a form field and a label of the form field;

determining category feature vectors based on the training model, each category feature vector indicating form features associated with a category in a quantity of categories;

determining form feature vectors based on the training model, each form feature vector indicating form features associated with a respective received form;

automatically associating, by a processor of a computing device, a set of forms with respective categories in the quantity of categories, using the training model and based on a similarity between the category feature vectors and form features associated with the set of forms;

automatically determining, based on form features of a particular form of the set of more forms, a combination of at least two categories to which the particular form belongs;

automatically adjusting the quantity of categories to include a combined category based on the combination of the at least two categories;

storing the forms and respective categories at an electronic form search server;

providing search results from the electronic form search server based on the combined category of the forms, wherein the search results include an identified form based on the form being associated with at least two categories; and

automatically modifying the training model to associate forms with respective categories in the adjusted quantity of categories.

10. The method of claim 9 , wherein creating the training model uses a collection of pre-categorized forms, the training model representing a significance of each of the form features to each of the categories.

11. The method of claim 9 , wherein creating the training model comprises using a machine learning algorithm on uncategorized forms in the collection of forms, the training model using coefficients representing belongingness to individual categories of a predetermined number of categories.

12. The method of claim 9 , wherein creating the training model comprises using a machine learning algorithm on uncategorized forms in the collection of forms, the training model using coefficients representing belongingness to individual categories of an adaptive number of categories.

13. The method of claim 9 further comprising tagging individual forms with a best matching category.

14. The method of claim 9 , wherein storing the forms and the respective categories comprises storing an indication of significance of a category to individual forms.

15. The method of claim 9 , wherein storing the forms and the respective categories comprises tagging individual forms with multiple relevant categories based on determining that the form is associated with more than one category.

16. A system comprising:

a processing device; and

a memory device communicatively coupled to the processing device,

wherein the processing device is configured to execute instructions included in the memory device to perform operations comprising:

receiving a quantity of categories, each category indicating a form feature, wherein the form features include (i) non-text field characteristics indicating a relation between multiple field locations, and (ii) field-specific text characteristics indicating an association between a form field and a label of the form field;

determining a category feature vector for each of the categories, each respective category feature vector indicating form features associated with the respective category;

determining a form feature vector for a set of forms, each respective form feature vector indicating form features associated with the respective form;

automatically associating, by the processing device, the set of forms with respective categories of the quantity of categories, the association based on a similarity between the category feature vectors of the respective categories and the form feature vectors of the set of forms;

automatically determining, based on form features of a particular form of the set of forms, a combination of at least two categories to which the particular form belongs;

automatically adjusting the quantity of categories to include a combined category based on the combination of the at least two categories;

storing the forms and the respective categories associated with the forms at an electronic form search server; and

providing search results from the electronic form search server based on input identifying the combined category, a form identified as a search result based on the form being associated with the at least two categories.

17. The system of claim 16 , wherein associating forms with respective categories comprises:

creating a training model using a collection of pre-categorized forms; and

applying the training model to additional forms to associate the additional forms with the respective categories based on form features for the non-text field characteristics or field-specific text characteristics of the additional forms.

18. The system of claim 16 , wherein associating forms with respective categories comprises:

using a machine learning algorithm to create a training model using a collection of pre-categorized forms, the training model using a respective coefficient representing a significance of each of the form features to each of the categories; and

applying the training model to additional forms to associate the additional forms with the respective categories based on form features for the non-text field characteristics or field-specific text characteristics of the additional forms.

19. The system of claim 16 , wherein associating forms with respective categories comprises:

receiving user input categorizing forms in a collection;

creating a training model using the collection of categorized forms; and

applying the training model to uncategorized forms to categorize the uncategorized forms in the respective categories based on the form features for the non-text field characteristics or field-specific text characteristics of the forms.

20. The system of claim 16 , wherein associating forms with respective categories comprising determining a feature vector for each form, wherein respective feature represent the form features, wherein associating the forms with the categories is based on the feature vectors.