| CPC G06Q 40/10 (2013.01) | 14 Claims |
|
1. A method for automatically recommending to self-employed users of a tax preparation software application one or more categories of a plurality of different categories of tax deductible expenses and examples of tax deductible expenses for each of the one or more categories, the method comprising:
providing a server comprising one or more processors for executing the tax preparation software, the server being in electronic communication with a computing device of each of the self-employed users and having access to a database of user information, wherein the computing device is configured to execute an application associated with the tax preparation software and is further configured to display an interface for interacting with the tax preparation software, wherein the user information comprises one or more prior tax returns filed by each respective self-employed user using the tax preparation software;
providing, by the server, first input data to a global binary classifier model, the first input data comprising a plurality of different first features associated with each of the self-employed users, wherein the plurality of different first features comprise a unique identifier for each respective self-employed user and information from the one or more prior tax returns filed by each respective self-employed user;
providing, by the server, second input data to the global binary classifier model, the second input data comprising a plurality of different second features associated with each of the self-employed users, wherein the plurality of different second features comprise an industry of each respective self-employed user, wherein the plurality of different second features further comprise a plurality of categories of tax deductible expenses, wherein each respective category of the plurality of categories is associated with examples of the tax deductible expenses and one or more industries that are associated with the respective category;
providing, by the server, third input data to the global binary classifier model, the third input data comprising, for each respective self-employed user, a plurality of tax deductible expenses selected and not selected by the respective self-employed user in the one or more prior tax returns, the third input data further comprising one or more contextual features for one or more of the self-employed users, the one or more contextual features comprising a recent life event associated with the one or more self-employed users;
training the global binary classifier model through a supervised learning process using the first input data, the second input data, and the third input data;
receiving, at the server, input data from the computing device of a first self-employed user of the plurality of self-employed users, the input data receive via user input provided via the interface of the computing device, the input data comprising the unique identifier for the first self-employed user, the industry for the first self-employed user, and the recent life event associated with the first self-employed user;
receiving, at the server, output from the global binary classifier model based, at least in part, on the input data, the output from the global binary classifier model comprising a recommendation for the first self-employed user, the recommendation comprising: (i) one or more categories of the plurality of different categories of tax deductible expenses; and (ii) a plurality of examples of tax deductible expenses for each of the one or more categories;
receiving, at the server, feedback from the first self-employed user on the recommendation, wherein the feedback comprises user selection or non-selection of one or more of the categories of tax deductible expenses;
generating, by the one or more processors, updated training data for the global binary classifier model based on the feedback; and
re-training the trained global binary classifier model in real-time based on the updated training data.
|