Apparatus for identifying root cause using unstructured data

摘要

A system and method of identifying root cause of an observation by leveraging features from unstructured data is disclosed. A report generation component may be configured to generate a report. A report presentation component may be configured to allow an operator to select an observation from the report. A root cause component may be configured to determine one or more causal factors associated with the observation.

主权项

1. A system for determining causal factors in unstructured data, the system comprising:
an unstructured data collection component configured to collect unstructured data from one or more sources of text; a natural language processing component configured to process text from the unstructured data to extract features from the unstructured data; a report generation component configured to generate a report based on features extracted from the unstructured data; a report presentation component configured to present the report to an operator and to allow the operator to select an observation from the report; a root cause component configured to determine one or more causal factors associated with the observation by calculating an impact of one or more of the features of the unstructured data on the observation selected by the operator using a baseline for comparison with the observation selected, the baseline being determined by the operator as either data comprising one or more features in which the observation is not present or data originating in a particular time period comprising one or more features in which the observation is present, and the impact on a measurable characteristic of the observation selected being calculated based on a comparison of one or more of the features of the unstructured data associated with the presence of the observation and features of the unstructured data associated with the baseline, the measurable characteristic being a volume-based metric, a sentiment metric, a satisfaction metric, or another user-defined metric, at least one of the one or more causal factors comprising one or more of the features; a memory configured to store the unstructured data collection component, the natural language processing component, the report generation component, the report presentation component, and the root cause component; and at least one processor to implement the unstructured data collection component, the natural language processing component, the report generation component, the report presentation component, and the root cause component.