System to assist users of a software application
AssigneeVoicemonk, Inc.
InventorsJagadeshwar Nomula, Vinesh Gudla
Life of this patent
Estimating the term…
Abstract
A system configured to assist users of a software application is provided. The system includes a virtual agent server 100 configured to receive input that identifies a primary user and at least one secondary user, wherein the primary user and the secondary user are parties to an event. Further, the virtual agent server 100 receives input identifying one or more preferences of the primary user and the secondary user corresponding to one or more categories of preferences. The virtual agent server 100 then assigns relative weightage to preferences within a category among categories of preferences. Subsequently, the virtual agent server 100 generates one or more suggestions corresponding to the event based on the relative weightage.
Claims20
A system comprising:
- a database comprising multimedia files, wherein one or more tags are associated with one or more of the multimedia files;
- and
- a virtual agent server, having a processor and a memory, configured to:
- receive input from a first user and a second user during a conversation between the first user and the second user;
- process the received input by applying natural language analysis to determine a context of the input;
- determine at least one preference of the first user and the second user based on the context, and generate a suggestion based on the at least one preference;
- identify one or more multimedia files from the database by matching the context to the tags;
- and
- share at least one of the identified multimedia files with at least one of the first user and the second user to enable better understanding between the first user and the second user,
- wherein the shared multimedia file corresponds to the generated suggestion,
- wherein the virtual agent server is configured to share the at least one of the identified multimedia files in the conversation between the first user and the second user through a device of at least one of the first user and the second user,
- wherein the at least one preference corresponds to one or more categories of preferences,
- wherein the virtual agent server is configured to assign relative weights to preferences within a category among the one or more categories of preferences based on a frequency of certain words used in the conversation;
- and
- wherein the virtual agent server is configured to generate a target response by applying a question-and-answer model;
- and wherein, to generate the suggestion, the virtual agent server is configured to:
- convert the determined at least one preference of the first user and the second user into a user vector;
- generate one or more suggestion results using a query that is constructed using the user vector;
- and
- rank the one or more suggestion results by using a Euclidean distance between a combination of the user vector and a query vector and a suggestion result vector, and
- re-rank the one or more ranked suggestion results by applying a gradient boosted decision tree learning model to features obtained from the user vector.
Dependencies are inferred from the claim text.