| 主权项 |
1. A computer-implemented method for improving a functionality of a computing device by optimizing improvements to an application running on the computing device, the computer-implemented method comprising:
creating, by one or more processors, a representational model of user reviews of an application by prioritizing defects in the application identified by the user reviews, wherein the representational model is trained and created exclusively with the user reviews of the application, and wherein the representational model is created by:
receiving, by one or more processors, user reviews of the application, wherein the user reviews comprise initial ratings of the application;classifying, by one or more processors, the user reviews according to a sentiment of the user reviews to identify negative user reviews and positive user reviews;classifying, by one or more processors, the negative user reviews into a review type selected from a group of review types consisting of a functionality type of review of the application, a performance type of review of the application, and a usability type of review of the application;tagging, by one or more processors, each of the negative user reviews with the review type;plotting, by one or more processors, the negative user reviews as negative user review representations in an n-dimensional space, wherein the negative user review representations are plotted according to the sentiment of the user reviews, the review type, and a contextual content of the negative user reviews;clustering, by one or more processors, top-k negative user review representations from the negative user review representations;plotting, by one or more processors, a defect in the application as a defect representation in the n-dimensional space;representing, by one or more processors, vectors from the defect representation to the top-k negative user review representations in the n-dimensional space, wherein the vectors represent a degree of overlap of contextual descriptors of the defect in the application and user complaints found in the negative user reviews;determining, by one or more processors, Euclidian distances from the defect representation to the top-k negative user review representations in the n-dimensional space according to the vectors;identifying, by one or more processors, the negative user review representations that are clustered within a predetermined distance from the defect representation according to the Euclidian distances; andprioritizing, by one or more processors, the defect in the application according to the negative user review representations that are clustered within the predetermined distance from the defect representation according to the Euclidian distances; utilizing the representational model to predict, by one or more processors, a rating improvement to the application caused by changing the application to resolve user complaints represented by the top-k negative user review representations that are clustered within the predetermined distance from the defect representation according to the Euclidian distances, wherein changes that resolve user complaints that are represented by the top-k negative user review representations that are clustered within the predetermined distance from the defect representation according to the Euclidian distances are predicted to result in an improvement to future user reviews of the application than changes that are represented by negative user review representations that are plotted outside the predetermined distance from the defect representation according to the Euclidian distances; based on the predicted rating improvement to the application, applying, by one or more processors, defect solutions that correct problems described in the top-k negative user review representations that are clustered within the predetermined distance from the defect representation according to the Euclidian distances to generate an improved version of the application; and installing, by one or more processors, the improved version of the application on the computing device to improve the functionality of the computing device. |
