Identifying suspicious entries in a document management system

1. A computer-implemented method of identifying suspicious entries in a document management system, the computer-implemented method comprising:

storing, by the document management system, a document including a plurality of entries of an entity associated with the document, each entry from the plurality of entries describing an activity at the entity at a given date and the entry added to the document by a respective employee of the entity;

storing, by the document management system, a security policy associated with the document, the security policy including one or more security rules defining attributes of suspicious entries in the document, a suspicious entry indicative of an activity at the entity described by the suspicious entry being potentially malicious, wherein the document management system is a machine learned model trained to generate at least one of the one or more security rules associated with the document;

receiving, by the document management system, a request by an employee of the entity to add a candidate entry to the document, the candidate entry describing an activity related to an object located at the entity and including a plurality of fields and a value for each of the plurality of fields that are associated with the activity described by the candidate entry,

wherein at least one value for one of the plurality of fields is provided by the employee of the entity in the request by the employee to add the candidate entry to the document;

classifying, by the document management system, the candidate entry as suspicious based on a comparison of a plurality of attributes of the candidate entry and the security policy prior to the candidate entry that is included in the request from the employee being added to the document or being rejected from being added to the document;

automatically adding, by the document management system, the candidate entry to the document responsive to the candidate entry being classified as suspicious despite the candidate entry being classified as suspicious;

transmitting, by the document management system, an alert that the candidate entry classified as suspicious was added to the document;

receiving, by the document management system, feedback on the classification of the candidate entry as suspicious, the feedback indicative of whether the classification is valid or invalid;

retraining the document management system that is the machine learned model based on the received feedback; and

generating, by the retrained document management system, one or more additional security rules associated with the document.

2. The computer-implemented method of claim 1 , further comprising:

receiving a definition for one of the one or more security rules from a client device, the definition including a plurality of attributes of malicious activity associated with the one of the one or more security rules.

3. The computer-implemented method of claim 1 , further comprising:

training the document management system using training data to generate the security rules for the document, the training data including suspicious entries and non-suspicious entries;

applying the document to the machine learned model; and

automatically generating, by the document management system, a security rule for the document based on the application of the document to the machine learned model.

4. The computer-implemented method of claim 3 , wherein the alert is transmitted to a client device of an administrator of the document management system and the feedback is from the client device of the administrator confirming or disputing the classification of the candidate entry as suspicious, the computer-implemented method further comprising:

preventing the candidate entry from being added to the document responsive to the administrator confirming the classification of the candidate entry as accurate; and

adding the candidate entry to the document responsive to the administrator disputing the classification of the candidate entry as suspicious.

5. The computer-implemented method of claim 4 , wherein the alert includes a list of a plurality of candidate entries classified as suspicious including the candidate entry classified as suspicious, the method further comprising:

receiving, for each of the plurality of candidate entries, feedback from the client device of the administrator including either a first indication confirming the classification of the candidate entry as accurate or a second indication disputing the classification of the candidate entry as suspicious.

6. The computer-implemented method of claim 1 , further comprising:

automatically rejecting the request to add the candidate entry to the document responsive to the candidate entry being classified as the suspicious entry.

7. The computer-implemented method of claim 1 , wherein classifying the candidate entry comprises:

extracting attributes of the candidate entry; and

comparing the extracted attributes of the candidate entry with the attributes of at least one of the one or more security rules included in the security policy;

wherein the candidate entry is classified as suspicious responsive to all of the extracted attributes of the candidate entry matching all of the attributes of the at least one of the one or more security rules, and the candidate entry is classified as non-suspicious responsive at least one of the extracted attributes of the candidate entry not matching at least one of the attributes of the at least one of the one or more security rules.

8. The computer-implemented method of claim 7 , further comprising:

automatically generating at least one security rule for entries assigned to a first destination in the document but the at least one security rule not applicable to entries assigned to a second destination in the document.

9. The computer-implemented method of claim 1 , wherein the attributes included in a security rule from the one or more security rules includes a destination in the document and at least one of a name of a representative of the entity, an invalid date range, an invalid time range, an invalid data value, or a location.

10. A non-transitory computer-readable storage medium storing executable code for identifying suspicious entries in a document management system, the code when executed by a computer processor causes the computer processor to perform steps including:

storing, by the document management system, a document including a plurality of entries of an entity associated with the document, each entry from the plurality of entries describing an activity at the entity at a given date and the entry added to the document by a respective employee of the entity;

storing, by the document management system, a security policy associated with the document, the security policy including one or more security rules defining attributes of suspicious entries in the document, a suspicious entry indicative of an activity at the entity described by the suspicious entry being potentially malicious, wherein the document management system is a machine learned model trained to generate at least one of the one or more security rules associated with the document;

receiving, by the document management system, a request by an employee of the entity to add a candidate entry to the document, the candidate entry describing an activity related to an object located at the entity and including a plurality of fields and a value for each of the plurality of fields that are associated with the activity described by the candidate entry, wherein at least one value for one of the plurality of fields is provided by the employee of the entity in the request by the employee to add the candidate entry to the document;

classifying, by the document management system, the candidate entry as suspicious based on a comparison of a plurality of attributes of the candidate entry and the security policy prior to the candidate entry that is included in the request from the employee being added to the document or being rejected from being added to the document;

automatically adding the candidate entry to the document responsive to the candidate entry being classified as suspicious despite the candidate entry being classified as suspicious;

transmitting an alert that the candidate entry classified as suspicious was added to the document;

receiving feedback on the classification of the candidate entry as suspicious, the feedback indicative of whether the classification is valid or invalid;

retraining the document management system that is the machine learned model based on the received feedback; and

generating, by the retrained document management system, one or more additional security rules associated with the document.

11. The non-transitory computer-readable storage medium of claim 10 , wherein the code when executed by the computer processor further causes the computer processor to perform steps including:

receiving a definition for one of the one or more security rules from a client device, the definition including a plurality of attributes of malicious activity associated with the one of the one or more security rules.

12. The non-transitory computer-readable storage medium of claim 10 , wherein the code when executed by the computer processor further causes the computer processor to perform steps including:

training the document management system using training data to generate the security rules for the document, the training data including suspicious entries and non-suspicious entries;

applying the document to the machine learned model; and

automatically generating a security rule for the document based on the application of the document to the machine learned model.

13. The non-transitory computer-readable storage medium of claim 12 , wherein the alert is transmitted to a client device of an administrator of the document management system and the feedback is from the client device of the administrator confirming or disputing the classification of the candidate entry as suspicious, the code when executed by the computer processor further causes the computer processor to perform steps including:

preventing the candidate entry from being added to the document responsive to the administrator confirming the classification of the candidate entry as accurate; and

adding the candidate entry to the document responsive to the administrator disputing the classification of the candidate entry as suspicious.

14. The non-transitory computer-readable storage medium of claim 13 , wherein the alert includes a list of a plurality of candidate entries classified as suspicious including the candidate entry classified as suspicious, the code when executed by the computer processor further causes the computer processor to perform steps including:

receiving, for each of the plurality of candidate entries, feedback from the client device of the administrator including either a first indication confirming the classification of the candidate entry as accurate or a second indication disputing the classification of the candidate entry as suspicious.

15. The non-transitory computer-readable storage medium of claim 12 , wherein classifying the candidate entry comprises:

extracting attributes of the candidate entry; and

comparing the extracted attributes of the candidate entry with the attributes of at least one of the one or more security rules included in the security policy; wherein the candidate entry is classified as suspicious responsive to all of the extracted attributes of the candidate entry matching all of the attributes of the at least one of the one or more security rules, and the candidate entry is classified as non-suspicious responsive at least one of the extracted attributes of the candidate entry not matching at least one of the attributes of the at least one of the one or more security rules.

16. The non-transitory computer-readable storage medium of claim 10 , wherein the attributes included in a security rule from the one or more security rules includes a destination in the document and at least one of a name of a representative of the entity, an invalid date range, an invalid time range, an invalid data value, or a location, wherein the code when executed by the computer processor further causes the computer processor to perform steps including:

automatically generating at least one security rule for entries assigned to a first destination in the document but the at least one security rule not applicable to entries assigned to a second destination in the document.

17. A document management system for identifying suspicious entries in the computer system, comprising:

one or more computer processors; and

a non-transitory computer-readable storage medium storing code, the code when executed by the one or more computer processors cause the one or more computer processors to perform steps comprising:

storing, by the document management system, a document including a plurality of entries of an entity associated with the document, each entry from the plurality of entries describing an activity at the entity at a given date and the entry added to the document by a respective employee of the entity;

storing, by the document management system, a security policy associated with the document, the security policy including one or more security rules defining attributes of suspicious entries in the document, a suspicious entry indicative of an activity at the entity described by the suspicious entry being potentially malicious, wherein the document management system is a machine learned model trained to generate at least one of the one or more security rules associated with the document;

receiving, by the document management system, a request by an employee of the entity to add a candidate entry to the document, the candidate entry describing an activity related to an object located at the entity and including a plurality of fields and a value for each of the plurality of fields that are associated with the activity described by the candidate entry, wherein at least one value for one of the plurality of fields is provided by the employee of the entity in the request by the employee to add the candidate entry to the document;

classifying, by the document management system, the candidate entry as suspicious based on a comparison of a plurality of attributes of the candidate entry and the security policy prior to the candidate entry that is included in the request from the employee being added to the document or being rejected from being added to the document;

automatically adding, by the document management system, the candidate entry to the document responsive to the candidate entry being classified as suspicious despite the candidate entry being classified as suspicious;

transmitting, by the document management system, an alert that the candidate entry classified as suspicious was added to the document;

receiving, by the document management system, feedback on the classification of the candidate entry as suspicious, the feedback indicative of whether the classification is valid or invalid;

retraining the document management system that is the machine learned model based on the received feedback; and

generating, by the retrained document management system, one or more additional security rules associated with the document.