Artificial bot detection for conferencing apps
One or more attendees joining a web conference are monitored. It is determined that one or more of the one or more attendees is an undesired attendee. In response to the determination, an action is caused to be performed.
1 . A system, comprising:
a communication interface; and
a processor coupled to the communication interface and configured to:
monitor one or more attendees joining a web conference, wherein the web conference is monitored using one or more web hooks;
generate embedding vectors in an embedding space using an identified set of attributes corresponding to each of the one or more attendees, wherein the identified set of attributes are generated using the one or more web hooks, and wherein the embedding space includes embeddings associated with a plurality of attendees attending a plurality of web conferences;
utilize, by a machine learning model, the embedding vectors to determine that a first attendee of the one or more attendees is an undesired attendee by determining that a corresponding embedding vector associated with the first attendee is less than a threshold distance to an undesired attendee cluster of the embedding space; and
in response to the determination by the machine learning model that the first attendee of the one or more attendees is the undesired attendee, provide to a host device an application programming interface (API) command to remove the first attendee.
2 . The system of claim 1 , wherein the communication interface is configured to receive an indication to monitor the web conference.
3 . The system of claim 1 , wherein the one or more attendees join the web conference using a corresponding device.
4 . The system of claim 3 , wherein the corresponding device is a computer, a tablet, a smartphone, a virtual machine, a container, a phone, a cell phone, or a smart device.
5 . The system of claim 1 , wherein the one or more attendees are monitored using the one or more web hooks.
6 . The system of claim 1 , wherein the one or more attendees are monitored using one or more application program interfaces (APIs).
7 . The system of claim 1 , wherein the identified set of corresponding attributes includes an attendee internet protocol (IP) address and associated details, a calling identifier or callback number if joining from a phone, attendee name, attendee profile image, attendee identity, and/or metadata.
8 . The system of claim 7 , wherein each attribute in the identified set of corresponding attributes is given an equal weight.
9 . The system of claim 7 , wherein at least one attribute in the corresponding set of attributes is given a weight that is different from other attributes included in the corresponding set of attributes.
10 . The system of claim 1 , wherein to determine that the one or more of the one or more attendees is the undesired attendee, the processor is configured to provide values associated with the identified set of corresponding attributes to one or more models.
11 . The system of claim 10 , wherein the one or more models include one or more machine learning models, one or more heuristic models, one or more statistical models, and/or one or more other mathematical models.
12 . The system of claim 10 , wherein the one or more models include one or more behavioral models configured to determine one or more behavioral attributes associated with the one or more attendees.
13 . The system of claim 12 , wherein the one or more behavioral attributes include whether the one or more attendees joined the web conference at a start time with second precision, whether the one or more attendees are participating in multiple meetings at a same time, and/or whether the one or more attendees have history of being silent in web conferences.
14 . The system of claim 1 , wherein in response to the determination by the machine learning model that the first attendee of the one or more attendees is the undesired attendee, the processor is further configured to provide a command to remove the undesired attendee from the web conference, provide a notification of a presence of the undesired attendee, notify the one or more attendees of the presence of the undesired attendee, cause the web conference to be terminated, and/or generate a security alert.
15 . A method, comprising:
monitoring one or more attendees joining a web conference, wherein the web conference is monitored using one or more web hooks;
generating embedding vectors in an embedding space using an identified set of attributes corresponding to each of the one or more attendees, wherein the identified set of attributes are generated using the one or more web hooks, and wherein the embedding space includes embeddings associated with a plurality of attendees attending a plurality of web conferences;
utilizing, by a machine learning model, the embedding vectors to determine that a first attendee of the one or more attendees is an undesired attendee by determining that a corresponding embedding vector associated with the first attendee is less than a threshold distance to an undesired attendee cluster of the embedding space; and
in response to the determination by the machine learning model that the first attendee of the one or more attendees is the undesired attendee, providing to a host device an application programming interface (API) command to remove the first attendee.
16 . The method of claim 15 , further comprising receiving an indication to monitor the web conference.
17 . The method of claim 15 , wherein the one or more attendees are monitored using the one or more web hooks and/or one or more APIs.
18 . The method of claim 15 , wherein in response to the determination by the machine learning model that the first attendee of the one or more attendees is the undesired attendee, the method further comprsises: providing a command to remove the undesired attendee from the web conference, providing a notification of a presence of the undesired attendee, notifying the one or more attendees of the presence of the undesired attendee, causing the web conference to be terminated, and/or generating a security alert.
19 . A computer program product embodied in a non-transitory computer readable medium and comprising computer instructions for:
monitoring one or more attendees joining a web conference, wherein the web conference is monitored using one or more web hooks;
generating embedding vectors in an embedding space using an identified set of attributes corresponding to each of the one or more attendees, wherein the identified set of attributes are generated using the one or more web hooks, and wherein the embedding space includes embeddings associated with a plurality of attendees attending a plurality of web conferences;
utilizing, by a machine learning model, the embedding vectors to determine that a first attendee of the one or more attendees is an undesired attendee by determining that a corresponding embedding vector associated with the first attendee is less than a threshold distance to an undesired attendee cluster of the embedding space; and
in response to the determination by the machine learning model that the first attendee of the one or more attendees is the undesired attendee, providing to a host device an application programming interface (API) command to remove the first attendee.