Sparse intent clustering through deep context encoders

Abstract

A method of sparse intent clustering is provided. The method comprises identifying features in a number of electronic user reports created by a user and contained in a database, wherein the features include a title and description. The features of each user report are encoded into a binary vector. The binary vector for each user report is fed into an autoencoder neural network that creates a N-dimensional vector representing the user report. The float vectors representing the user reports are projected into a N-dimensional space. The float vectors are clustered according to cosine similarities, wherein each vector cluster represents an intent of the user in creating the reports. The intent of each vector cluster is then labeled.

1 .- 20 . (canceled) 
     
     
         21 . A system, comprising:
 one or more processors coupled with memory, the one or more processors configured to:   receive, from a client device, an instruction to create an electronic report based on a title and a description;   determine an intent of the electronic report based on the title and the description;   identify a vector cluster of a plurality of vector clusters having a labeled intent that matches the intent determined for the electronic report, wherein the plurality of vector clusters is grouped according to cosine similarities between vectors generated via a neural network from features of electronic reports stored in a database; and   provide, for display via the client device, one or more features from the vector cluster for inclusion in the electronic report.   
     
     
         22 . The system of  claim 21 , wherein the one or more processors are further configured to:
 generate the plurality of vector clusters from the electronic reports previously generated in response to instructions from the client device that provides the instruction to create the electronic report.   
     
     
         23 . The system of  claim 21 , wherein the one or more processors are further configured to:
 generate, via the neural network, the vectors comprising compressed float vectors.   
     
     
         24 . The system of  claim 21 , comprising wherein the one or more processors are further configured to:
 generate the vectors via the neural network comprising an autoencoder neural network.   
     
     
         25 . The system of  claim 21 , wherein the one or more processors are further configured to:
 generate the vectors based on one or more multi-dimensional float vectors comprising floating point numbers with fractional parts that represent the electronic reports stored in the database.   
     
     
         26 . The system of  claim 21 , wherein the one or more processors are further configured to:
 encode the features of the electronic reports stored in the database into one or more binary vectors; and   input the one or more binary vectors into the neural network to generate the vectors.   
     
     
         27 . The system of  claim 26 , wherein the one or more processors are further configured to:
 encode the features into the one or more binary vectors via one-hot encoding a number of fields corresponding to the features.   
     
     
         28 . The system of  claim 21 , wherein the one or more processors are further configured to:
 cluster the vectors into the plurality of vector clusters based on the cosine similarities being greater than or equal to a threshold.   
     
     
         29 . The system of  claim 21 , wherein the one or more processors are further configured to:
 project the vectors into a multi-dimensional space having a predetermined number of dimensions; and   cluster the vectors projected into the multi-dimensional space according to the cosine similarities to create the plurality of vector clusters.   
     
     
         30 . The system of  claim 21 , wherein the one or more features comprise at least one of a field or a filter. 
     
     
         31 . The system of  claim 21 , wherein the one or more processors are further configured to:
 receive, from the client device, a selection of a first feature from the one or more features; and   create the electronic report with the first feature that is selected.   
     
     
         32 . The system of  claim 21 , wherein the one or more processors are further configured to:
 generate the plurality of vector clusters from the electronic reports previously generated by a same user of the client device in response to the instruction to create the electronic report.   
     
     
         33 . A method, comprising:
 receiving, by one or more processors coupled with memory, from a client device, an instruction to create an electronic report based on a title and a description;   determining, by the one or more processors, an intent of the electronic report based on the title and the description;   identifying, by the one or more processors, a vector cluster of a plurality of vector clusters having a labeled intent that matches the intent determined for the electronic report, wherein the plurality of vector clusters is grouped according to cosine similarities between vectors generated via a neural network from features of electronic reports stored in a database; and   providing, by the one or more processors for display via the client device, one or more features from the vector cluster for inclusion in the electronic report.   
     
     
         34 . The method of  claim 33 , comprising:
 generating, by the one or more processors, the plurality of vector clusters from the electronic reports previously generated in response to instructions from the client device that provides the instruction to create the electronic report.   
     
     
         35 . The method of  claim 33 , comprising:
 generating, by the one or more processors via the neural network, the vectors comprising compressed float vectors.   
     
     
         36 . The method of  claim 33 , comprising:
 generating, by the one or more processors, the vectors via the neural network comprising an autoencoder neural network.   
     
     
         37 . The method of  claim 33 , comprising:
 generating, by the one or more processors, the vectors based on one or more multi-dimensional float vectors comprising floating point numbers with fractional parts that represent the electronic reports stored in the database.   
     
     
         38 . The method of  claim 33 , comprising:
 encoding, by the one or more processors, the features of the electronic reports stored in the database into one or more binary vectors; and   inputting, by the one or more processors, the one or more binary vectors into the neural network to generate the vectors.   
     
     
         39 . A non-transitory computer-readable medium storing processor executable instructions that, when executed by one or more processors, cause the one or more processors to:
 receive, from a client device, an instruction to create an electronic report based on a title and a description;   determine an intent of the electronic report based on the title and the description;   identify a vector cluster of a plurality of vector clusters having a labeled intent that matches the intent determined for the electronic report, wherein the plurality of vector clusters is grouped according to cosine similarities between vectors generated via a neural network from features of electronic reports stored in a database; and   provide, for display via the client device, one or more features from the vector cluster for inclusion in the electronic report.   
     
     
         40 . The non-transitory computer-readable medium of  claim 39 , wherein the instructions further include instructions to:
 generate the plurality of vector clusters from the electronic reports previously generated in response to instructions from the client device that provides the instruction to create the electronic report.