Semantic Paths: Connecting research at EMNLP 2021

Semantic Paths brings together the power of Babelscape's neuro-symbolic multilingual text processing which includes state-of-the-art Word Sense Disambiguation and Entity Linking, concept extraction and personalized knowledge graph creation.

For the EMNLP 2021 conference use case, we applied our technologies to the document base of the EMNLP 2021 accepted papers (including Findings of EMNLP), to extract meaningful insights and correlations from raw data between papers, researchers, research bodies, skills, areas, etc.:

• We created a domain-specific knowledge graph containing thousands of NLP-related concepts and Named Entities interlinked semantically.
• Babelscape’s multilingual text analytics tools such as Extraggo and Comprehendo were used to extract semantically-enhanced gists of articles texts.
• Correlations between user queries and knowledge graph entities (researchers, research areas, skills, etc.) were computed as the outcome of our state-of-the-art Word Sense Disambiguation and entity linking system.
• The EMNLP 2021 authors collection was enriched with a selection of authoritative researchers in the NLP field.
• Researchers’ profiles and affiliations were obtained by integrating information from various open data sources.

SemanticPaths can be customized to search any domain or document base. Find out more here.