Utilizing similarity, knowledge graphs, and machine learning models to refine/redefine a target skill profile based on skills data from employees of an entity and activate a recommendation system

A device, comprising:

• one or more memories;
• and
• one or more processors, communicatively coupled to the one or more memories, configured to:
• receive employee data identifying one or more of:
• skills,
• completed trainings,
• organizations,
• interests,
• expertise, or
• an education level associated with an employee of one or more employees;
• determine, from the employee data, skill data identifying skills of the employee;
• obtain historical data,
• wherein the historical data is associated with at least one of:
• historical employee data,
• historical skill data,
• historical similarity scores, and/or
• historical groups of anchor skill data;
• obtain, based on the historical data, a set of observations;
• partition the set of observations into a training set;
• train a similarity model and/or a clustering model with the training set;
• process the employee data and the skill data, with the similarity model, to determine similarity scores between the skills identified by the skill data based on the employee data,
• wherein the similarity model includes a cosine similarity on a weighted bipartite graph model;
• add or remove one or more skills to or from the skill data for predefined target skill profile categories, based on the similarity scores and to generate modified skill data,
• wherein the predefined target skill profile categories are based on an expert-defined hierarchy of target skill profiles and skills for an entity,
• wherein the expert-defined hierarchy of target skill profiles and skills for the entity includes:
• a first level corresponding to the entity,
• a second level corresponding to one or more predefined target skill profile categories, and
• a third level corresponding to one or more skills associated with a predefined target skill profile category of the one or more predefined target skill profile categories;
• compare the similarity scores and a predefined threshold to determine anchor skill data from the modified skill data and for each of the predefined target skill profile categories,
• wherein the one or more processors, to compare the similarity scores and the predefined threshold to determine the anchor skill data, are configured to:
• initially define a set of skills having a total similarity score above the predefined threshold as anchor skills,
• selectively add or remove a skill, of the set of skills, based on comparing a weight of the skill, calculated relative to the predefined target skill profile category, to a particular threshold from one or more seeds of the third level, and
• redefine the expert-defined hierarchy based on selecting, adding or removing the skill and based on performing an iterative process to continue to add new skills, of the set of skills, as seeds to the one or more seeds in the third level;
• group the anchor skill data for corresponding pluralities of the predefined target skill profile categories, based on a predefined group category and to generate groups of the anchor skill data;
• process the groups of the anchor skill data and the similarity scores, with the clustering model, to generate clustered anchor skill data for each of the groups of the anchor skill data,
• wherein the clustering model includes a hierarchical clustering model that recursively merges communities into a single node and executes modularity clustering on condensed graphs using the similarity scores to generate resulting clusters, and
• wherein the clustering model defines a resulting cluster, of the resulting clusters, that contain less than a particular quantity of skills as a unique cluster to reduce granularity of the clustered anchor skill data;
• identify one or more job opportunities for the employee based on the clustered anchor skill data;
• and
• provide data identifying the one or more job opportunities to a client device associated with the employee.

A non-transitory computer-readable medium storing instructions, the instructions comprising:

• one or more instructions that, when executed by one or more processors, cause the one or more processors to:
• receive employee data identifying one or more of:
• self-assessed skills,
• completed trainings,
• organizations,
• interests,
• expertise, or
• an education level associated with an employee;
• determine, from the employee data, skill data identifying skills of the employee;
• obtain historical data,
• wherein the historical data is associated with at least one of:
• historical employee data,
• historical skill data,
• historical similarity scores, and/or
• historical groups of anchor skill data;
• obtain, based on the historical data, a set of observations;
• partition the set of observations into a training set;
• train a similarity model and/or a clustering model with the training set;
• process the employee data and the skill data, with a similarity model, to determine similarity scores between the skills identified by the skill data based on the employee data,
• wherein the similarity model includes a cosine similarity on a weighted bipartite graph model;
• add or remove one or more skills to or from the skill data for predefined target skill profile categories, based on the similarity scores and to generate modified skill data,
• wherein the predefined target skill profile categories are based on an expert-defined hierarchy of target skill profiles and skills for an entity,
• wherein the expert-defined hierarchy of target skill profiles and skills for the entity includes:
• a first level corresponding to the entity,
• a second level corresponding to one or more predefined target skill profile categories, and
• a third level corresponding to one or more skills associated with a predefined target skill profile category of the one or more predefined target skill profile categories;
• compare the similarity scores and a predefined threshold to determine anchor skill data from the modified skill data and for each of the predefined target skill profile categories,
• wherein the one or more instructions, that cause the one or more processors to compare the similarity scores and the predefined threshold to determine the anchor skill data, cause the one or more processors to to:
• initially define a set of skills having a total similarity score above the predefined threshold as anchor skills,
• selectively add or remove a skill, of the set of skills, based on comparing a weight of the skill, calculated relative to the predefined target skill profile category, to a particular threshold from one or more seeds of the third level, and
• redefine the expert-defined hierarchy based on selecting, adding or removing the skill and based on performing an iterative process to continue to add new skills, of the set of skills, as seeds to the one or more seeds in the third level;
• group the anchor skill data for corresponding pluralities of the predefined target skill profile categories, based on a predefined group category and to generate groups of the anchor skill data;
• process the groups of the anchor skill data and the similarity scores, with the clustering model, to generate clustered anchor skill data for each of the groups of the anchor skill data,
• wherein the clustering model includes a hierarchical clustering model that recursively merges communities into a single node and executes modularity clustering on condensed graphs using the similarity scores to generate resulting clusters, and
• wherein the clustering model defines a resulting cluster, of the resulting clusters, that contain less than a particular quantity of skills as a unique cluster to reduce granularity of the clustered anchor skill data;
• and
• perform one or more actions based on the clustered anchor skill data.