CPC G06F 16/285 (2019.01) [G06F 16/243 (2019.01); G06F 16/24578 (2019.01); G06Q 30/0206 (2013.01)] | 18 Claims |
1. A computer-implemented method comprising:
receiving, by a server, opportunity data that describes an opportunity and unstructured supporting data that is in multiple different data formats that are not consumable by a set of trained interconnected classifiers, that has been collected from a plurality of different data sources, and that is for aiding the server in generating a plurality of data insights for improving a potential for winning the opportunity;
obtaining, by the server, unstructured historical data from a data repository based on the received data that describes the opportunity;
generating, by the server and based on the received, unstructured supporting data and the obtained, unstructured historical data, standardized feature vectors that are in a standardized data format that is consumable by the set of trained interconnected classifiers, wherein generating standardized feature vectors comprises:
tokenizing the received, unstructured supporting data and the obtained, unstructured historical data to identify multiple tokens, and
transforming the multiple tokens into the standardized feature vectors in an N-dimensional data set;
filtering, by the server, the standardized feature vectors based at least on the opportunity data to obtain filtered standardized feature vectors with a reduced dimensionality, wherein filtering the standardized feature vectors comprises: for each feature vector included in the standardized feature vectors,
comparing i) text corresponding to the feature vector with ii) text of the opportunity data,
determining that the text of the feature vector is not included in the text of the opportunity data, and
in response to determining that the text of the feature vector is not included in the text of the opportunity data, removing the feature vector, whose text was not included in the text of the opportunity data, from the standardized feature vectors;
generating, by the server, a win percentage for the opportunity based on the filtered standardized feature vectors using a particular trained interconnected classifier of the set of trained interconnected classifiers;
in response to generating the win percentage for the opportunity, generating, by the server and for each remaining trained interconnected classifier of the set of trained interconnected classifiers, a different data insight that is unique to the trained interconnected classifier; and
providing, by the server, the win percentage and the plurality of data insights regarding the opportunity for output.
|