Joint Representation Learning for Social Network Talent Search Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional querying methods for talent searches on social networks are inefficient, requiring manual entry of search terms and extensive trial-and-error to identify suitable candidates, as they need to translate complex job position criteria into effective search queries, which is challenging due to the dynamic nature of required skills and knowledge.
Innovation Solution
A system that represents standardized entities like skills, titles, and locations in a heterogeneous graph, using node proximity to calculate affinity scores and train embeddings for search query terms, allowing for the recommendation of refined search terms and facets to improve query accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual search term entry is used, then recruiters can perform talent searches, but the effort and time required to formulate effective queries increases significantly
Solution Approach 1:
The system enables self-service by automatically generating search queries and suggesting search terms based on job position criteria, eliminating the need for recruiters to manually formulate complex queries. The patent implements this through automated query generation that translates position requirements into effective search queries without manual intervention.
Solution Approach 2:
The patent introduces an intermediary system that acts as a bridge between job position criteria and effective search queries. This intermediary component automatically translates hiring position requirements into optimized search queries, reducing the manual effort and time recruiters need to invest in query formulation.
2Measurement precision
If standardized entities are represented in a heterogeneous graph with machine learning models, then query accuracy improves, but system complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the complex task of query optimization into distinct modules: a heterogeneous graph structure for representing entities and relationships, embedding generation for converting entities to vector representations, and affinity score calculation for measuring relevance. This modular segmentation manages system complexity while maintaining high query accuracy.
Solution Approach 2:
The system utilizes parameter changes by transforming standardized entities into embedding vectors in a multi-dimensional space, where the parameters (dimensions) capture semantic relationships. This transformation enables accurate affinity score calculation and improves query precision without proportionally increasing operational complexity.
Data Source
AI summary
An indication of a plurality of different entities in a social networking service is received, including at least two entities having a different entity type. A plurality of user profiles in the social networking service is accessed. A first machine-learned model is used to learn embeddings for the plurality of different entities in a d-dimensional space. A second machine-learned model is used to learn an embedding for each of one or more query terms that are not contained in the indication of the plurality of different entities in the social networking service, using the embeddings for the plurality of different entities learned using the first machine-learned model, the second-machine learned model being a deep structured semantic model (DSSM). A similarity score between a query term and an entity is calculated by computing distance between the embedding for the query term and the embedding for the entity in the d-dimensional space.