System and method for the detection and visualization of reported ethics cases within an organization

A method for the detection and visualization of reported ethics cases within an organization comprising:

• receiving a set of digital records corresponding to reported ethics violations as corresponding reported claims in a textual form from a plurality of disparate systems;
• converting, each of the digital records from the set of digital records into a common digital format, wherein the common digital format comprises uniform text structure;
• deconstructing the uniform text structure of each digital record by a natural language processing module to lemmatize words, remove punctuation, and remove stop words from each of the digital records;
• receiving the deconstructed uniform text structure of each of the digital records from the natural language processing module;
• inputting each deconstructed uniform text structure into a binary machine learning data model, wherein the binary machine learning data model returns a determination value for the digital record corresponding to each given one of the reported claims, wherein the determination value consists of a binary indication as to whether the given one of the reported claims comprises an ethical allegation or not;
• inputting each deconstructed uniform text structure into a multiclass machine learning data model, wherein the multiclass machine learning data model returns a determination of a most likely allegation class for the given one of the reported claims when the given one of the reported claims comprises an ethical allegation;
• inputting each deconstructed uniform text structure into a multi-label machine learning model, wherein the multi-label machine learning model returns a determination of additional allegation classes for the given one of the reported claims when the given one of the reported claims pertains to more than one allegation class;
• inputting the determined value, the most likely allegation class, and the additional allegation classes to an ensemble machine learning data model, wherein the ensemble machine learning data model returns a determination of a possible class corresponding to each given one of the reported claims, wherein the possible class belongs to one of a plurality of available classes of ethics violations including alcohol and drug violations, corruption violations, conflict of interest violations, discrimination violations, failure to report violations, financial integrity violations, harassment violations, insider trading violations, retaliation violations, and workplace violence violations;
• prioritizing reported claims into one or more lists based on the determination of the possible class;
• and
• transmitting the prioritized list to a display;
• wherein at least one of the learning data models is trained using the deconstructed uniform text structure of the digital records as training input, such that the at least one of the learning data models is trained using the reported claims in a normalized format where specialized characters, punctuation, and stop words are removed and remaining words are lemmatized.

A non-transitory computer readable medium, having stored thereon, instructions that when executed by a computing device, cause the computing device to perform operations comprising:

• receiving a set of digital records corresponding to reported ethics violations as corresponding reported claims in a textual form from a plurality of disparate systems;
• converting, each of the digital records from the set of digital records into a common digital format, wherein the common digital format comprises uniform text structure;
• deconstructing the uniform text structure of each digital record by a natural language processing module to lemmatize words, remove punctuation, and remove stop words from each of the digital records;
• receiving the deconstructed uniform text structure of each of the digital records from the natural language processing module;
• inputting each deconstructed uniform text structure into a binary machine learning data model, wherein the binary machine learning data model returns a determination value for the digital record corresponding to each given one of the reported claims, wherein the determination value consists of a binary indication as to whether the given one of the reported claims comprises an ethical allegation or not;
• inputting each deconstructed uniform text structure into a multiclass machine learning data model, wherein the multiclass machine learning data model returns a determination of a most likely allegation class for the given one of the reported claims when the given one of the reported claims comprises an ethical allegation;
• inputting each deconstructed uniform text structure into a multi-label machine learning model, wherein the multi-label machine learning model returns a determination of additional allegation classes for the given one of the reported claims when the given one of the reported claims pertains to more than one allegation class;
• inputting the determined value, the most likely allegation class, and the additional allegation classes to an ensemble machine learning data model, wherein the ensemble machine learning data model returns a determination of a possible class corresponding to each given one of the reported claims, wherein the possible class belongs to one of a plurality of available classes of ethics violations including alcohol and drug violations, corruption violations, conflict of interest violations, discrimination violations, failure to report violations, financial integrity violations, harassment violations, insider trading violations, retaliation violations, and workplace violence violations;
• prioritizing reported claims into one or more lists based on the determination of the possible class;
• and
• transmitting the prioritized list to a display;
• wherein at least one of the learning data models is trained using the deconstructed uniform text structure of the digital records as training input, such that the at least one of the learning data models is trained using the reported claims in a normalized format where specialized characters, punctuation, and stop words are removed and remaining words are lemmatized.