Restricted access to sensitive content
In one aspect, the present disclosure relates to a method including: receiving, by a client device, a request to access content stored on a remote server; determining, by the client device, that the requested content includes sensitive information based on a user profile associated with the client device; modifying, by the client device, the requested content in response to the determination that the content includes sensitive information; and providing, by client device, access to the modified content in place of the requested content that includes the sensitive information.
1. A method comprising:
detecting, by a client device, a request to access content located on a remote server;
determining, by the client device, that the requested content includes sensitive information based on a profile associated with the client device and a context in which the content is being accessed, comprising:
extracting text from the requested content;
generating one or more feature vectors based on the extracted text; and
using the one or more feature vectors and one or more trained models to detect the sensitive information of the requested content, wherein a training of the one or more models comprise:
receiving a training dataset comprising one or more sensitive documents and one or more non-sensitive documents;
for individual documents in the training dataset, extracting tokens from text of the document, performing a Term Frequency-Inverse Document Frequency (TF-IDF) calculation using the tokens, and generating feature vectors based on the TF-IDF calculation; and
using the feature vectors generated for individual documents in the training dataset to generate the one or more trained models;
modifying, by the client device, the requested content in response to the determination that the content includes sensitive information; and
providing, by client device, access to the modified content in place of the requested content that includes the sensitive information.
2. The method of claim 1 , wherein the determination that the requested content includes sensitive information is further based on metadata associated with the content.
3. The method of claim 1 , wherein the determination that the requested content includes sensitive information further includes analysis of the requested content using the one or more trained models selected based on the user profile.
4. The method of claim 3 , further including selecting the one or more trained models based on attributes of an organization with which the requested content is associated.
5. The method of claim 1 , wherein the extraction of the text from the requested content includes using Optical Character Recognition (OCR) to extract the text.
6. The method of claim 1 , wherein the detection of the access to the requested content located on the remote server and the determination the requested content includes sensitive information is performed by a browser application of the client device.
7. The method of claim 6 , wherein the requested content is located within a Software-as-a-Service (SaaS) application.
8. The method of claim 1 , wherein the modification of the requested content includes a change to a document using an application programming interface (API), and the API being selected based on a file type of the content.
9. The method of claim 1 , wherein the request to access the content includes a request to upload, download, share, copy, or paste the content.
10. The method of claim 1 , wherein the performing of the TF-IDF calculation using the tokens comprises:
calculating a Document Frequency (DF) value for each of the tokens; and
determining which of the tokens to include in the feature vectors for the individual documents in the training dataset based on the calculated DF value for each of the tokens.
11. A device comprising:
a memory; and
a processor coupled to the memory and configured to:
access content of an application, the content including sensitive information, and the application being executable on a remote computing device;
detect the sensitive information of the content based on at least one of a user profile and a context in which the content is being accessed, comprising:
extract text from the requested content;
generate one or more feature vectors based on the extracted text; and
using the one or more feature vectors and one or more trained models to detect the sensitive information of the requested content, wherein a training of the one or more models comprise:
receive a training dataset comprising one or more sensitive documents and one or more non-sensitive documents;
for individual documents in the training dataset, extract tokens from text of the document, perform a Term Frequency-Inverse Document Frequency (TF-IDF) calculation using the tokens, and generate feature vectors based on the TF-IDF calculation; and
using the feature vectors generated for individual documents in the training dataset to generate the one or more trained models; and
modify the content in response to detection of the sensitive information, the modification enabling the computing device to replicate security controls applicable to local users of the application.
12. The device of claim 11 , wherein the processor is configured to detect the sensitive information of the content further based on metadata associated with the content.
13. The device of claim 11 , wherein the processor is configured to detect the sensitive information of the content by analyzing the content using the one or more trained models selected based on at least one of the user profile and the context in which the content is being accessed.
14. The device of claim 13 , wherein the processor is configured to select the one or more trained models based on attributes of an organization with which the content is associated.
15. The device of claim 11 , wherein the processor is configured to use Optical Character Recognition (OCR) to extract the text.
16. The device of claim 11 , further including a browser application that, when executed by the processor, is operable to detect the sensitive information of the content and to modify the content in response to detection of the sensitive information.
17. The device of claim 16 wherein the application is a Software-as-a-Service (SaaS) application.
18. The device of claim 11 , wherein the processor is configured to:
calculate a Document Frequency (DF) value for each of the tokens; and
determine which of the tokens to include in the feature vectors for the individual documents in the training dataset based on the calculated DF value for each of the tokens.
19. A method comprising:
detecting, by a gateway device, a request to access content stored on a remote server, the request being associated with a client device;
determining, by the gateway device, that the requested content includes sensitive information based on a user profile associated with the client device, comprising:
extracting text from the requested content;
generating one or more feature vectors based on the extracted text; and
using the one or more feature vectors and one or more trained models to detect the sensitive information of the requested content, wherein a training of the one or more models comprise:
receiving a training dataset comprising one or more sensitive documents and one or more non-sensitive documents;
for individual documents in the training dataset, extracting tokens from text of the document, performing a Term Frequency-Inverse Document Frequency (TF-IDF) calculation using the tokens, and generating feature vectors based on the TF-IDF calculation; and
using the feature vectors generated for individual documents in the training dataset to generate the one or more trained models;
modifying, by the gateway device, the requested content in response to the determination that the content includes sensitive information; and
providing, by gateway device, access to the modified content in place of the requested content that includes the sensitive information.
20. The method of claim 19 , wherein the determination that the requested content includes sensitive information further includes analysis of the requested content using the one or more trained models selected based on the user profile associated with the client device.