Machine learning techniques for detecting docketing data anomalies

1. A method for automatically detecting docketing data anomalies, the method comprising:

storing, in a docketing system, docketing information for a plurality of matters, each of the plurality of matters including a plurality of activities and a plurality of documents:

receiving a first document of the plurality of documents;

identifying one or more docketing activities, associated with the first matter, that are generated in response to receiving the first document;

retrieving the first document from the plurality of documents associated with a first matter of the plurality of matters;

determining a document type of the first document;

extracting one or more features from the first document and the plurality of activities associated with the first matter, the one or more features that are extracted comprising the one or more docketing activities and a title of the first document; and

training a machine learning model, based on the extracted features and the document type of the first document, to determine one or more expected docketing activities for a new document determined to match the document type.

2. The method of claim 1 , further comprising identifying one or more deadlines, associated with the first matter, that are generated in response to receiving the first document, wherein the one or more features include the one or more deadlines.

3. The method of claim 1 further comprising determining a sequence of the one or more activities, wherein the one or more features include the sequence of the one or more activities.

4. The method of claim 1 further comprising determining a status or age of the first matter, wherein the age is determined based on a filing date associated with the first flatter, and wherein the one or more features include the status or age of the first matter.

5. The method of claim 1 further comprising:

receiving a second document associated with a second matter of the plurality of matters;

obtaining a list of docketing activities generated for the second matter in response to receiving the second document;

applying the trained machine learning model to the second document to determine one or more expected docketing activities for the second document; and

detecting a docketing anomaly in response to determining that the obtained list of docketing activities does not match the one or more expected docketing activities for the second document.

6. The method of claim 1 further comprising:

training the machine learning model, based on the one or more features extracted from the first document, to classify a country associated with the first document.

7. The method of claim 6 , wherein the classifying the country associated with the first document comprises classifying the first document as a Patent Cooperation Treaty (PCT) document or other document.

8. The method of claim 6 further comprising:

receiving a second document;

determining that the second document has been associated with a given country;

applying the trained machine learning model to the second document to determine an expected country for the second document; and

detecting a docketing anomaly in response to determining that the given country associated with the second document does not match the expected country for the second document.

9. The method of claim 1 , wherein the first document includes a plurality of checkboxes one of which is checked, and wherein the one or more features include a field for each checkbox, an indication of the field corresponding to the checkbox that is checked, a list of previous activities that predate the first document, and a given expected activity that is associated with the field corresponding to the checkbox that is checked and the list of previous activities.

10. A patent management system comprising:

a server, operatively connected to a network, wherein the server includes:

a processor; and

a memory; and

the processor configured to perform operations comprising:

storing in a docketing system docketing information for a plurality of matters, each of the plurality of matters including a plurality of activities and a plurality of documents;

receiving a first document of the plurality of documents;

identifying one or more docketing activities, associated with the first matter, that are generated in response to receiving the first document;

retrieving a first document from the plurality of documents associated with a first matter of the plurality of matters;

determining a document type of the first document;

extracting one or more features from the first document and the plurality of activities associated with the first matter, the one or more features that are extracted comprising the one or more docketing activities and a title of the first document; and

training a machine learning model, based on the extracted features and the document type of the first document, to determine one or more expected docketing activities for a new document determined to match the document type.

11. The patent management system of claim 10 , wherein the processor is further configured to identify one or more deadlines, associated with the first matter, that are generated in response to receiving the first document, wherein the one or more features include the one or more deadlines.

12. The patent management system of claim 10 , wherein the processor is further configured to determine a sequence of the one or more activities, wherein the one or more features include the sequence of the one or more activities.

13. The patent management system of claim 10 , wherein the processor is further configured to determine a status or age of the first matter, wherein the age is determined based on a filing date associated with the first matter, and wherein the one or more features include the status or age of the first matter.

14. The patent management system of claim 10 , wherein the processor is further configured to:

receive a second document associated with a second matter of the plurality of matters;

obtain a list of docketing activities generated for the second matter in response to receiving the second document;

apply the trained machine learning model to the second document to determine one or more expected docketing activities for the second document; and

detect a docketing anomaly in response to determining that the obtained list of docketing activities does not match the one or more expected docketing activities for the second document.

15. The patent management system of claim 10 , wherein the processor is further configured to train the machine learning model; based on the one or more features extracted from the first document, to classify a country associated with the first document.

16. The patent management system of claim 15 , wherein the classifying the country associated with the first document comprises classifying the first document as a Patent Cooperation Treaty (PCT) document or other document.

17. The patent management system of claim 15 , wherein the processor is further configured to:

receive a second document;

determine that the second document has been associated with a given country;

apply the trained machine learning model to the second document to determine an expected country for the second document; and

detect a docketing anomaly in response to determining that the given country associated with the second document does not match the expected country for the second document.

18. A non-transitory computer-readable medium comprising non-transitory computer readable instructions that, when executed by one or more processors, configure the one or more processors to perform operations comprising:

storing, in a docketing system, docketing information for a plurality of matters, each of the plurality of matters including a plurality of activities and a plurality of documents;

receiving a first document of the plurality of documents;

identifying one or more docketing activities, associated with the first matter, that are generated in response to receiving the first document;

retrieving the first document from the plurality of documents associated with a first matter of the plurality of matters;

determining a document type of the first document;

extracting one or more features from the first document and the plurality of activities associated with the first matter, the one or more features that are extracted comprising the one or more docketing activities and a title of the first document; and

training a machine learning model, based on the extracted features and the document type of the first document, to determine one or more expected docketing activities for a new document determined to match the document type.