Knowledge Tree
semantic similarity
Also known as: semantic similarity metrics, semantic similarity measures, semantic similarity search
Facts (31)
Sources
Applying Large Language Models in Knowledge Graph-based ... arxiv.org Jan 7, 2025 7 facts
claimThe integration of two concepts within a knowledge graph is frequently described as a semantic mapping process based on approaches for elaborating semantic similarity.
claimSemantic similarity in knowledge graphs is developed by introducing quantitative values to the relationship between two concepts.
claimSemi-automatic processing approaches for mapping tasks utilize a combination of automated semantic similarity measures and human expert experience.
referenceZhu, G. and Iglesias, C.A. published the paper 'Computing semantic similarity of concepts in knowledge graphs' in the IEEE Transactions on Knowledge and Data Engineering in 2016.
claimKnowledge-based approaches assess semantic similarity by incorporating the hierarchy of concepts within a graph, such as identifying the least common subsumer of two concepts.
claimCorpus-based approaches for determining semantic similarity between concepts rely on data from extensive corpora like Wikipedia and measure semantic relatedness rather than semantic similarity, meaning they do not account for hierarchical relations.
claimLLM-based and KG-based approaches use knowledge graphs as input, but LLM-based approaches shift the processing methodology away from semantic similarity measures toward using LLMs to assess domain concept instantiation within a modeling language.
Knowledge Graphs: Opportunities and Challenges - Springer Nature link.springer.com Apr 3, 2023 5 facts
referenceWordNet is a lexical knowledge base that uses hierarchical concept graphs to analyze semantic similarity between words.
claimEntity disambiguation methods that rely on rich contextual information fail to precisely measure the semantic similarity of entities when the source texts are short and have limited context.
claimZhu and Iglesias (2018) proposed the SCSNED method for entity disambiguation, which measures semantic similarity based on both informative words of entities in knowledge graphs and contextual information found in short texts.
claimKnowledge graph-based information retrieval offers the advantage of semantic representation of items, where items are represented via a formal and interlinked model that supports semantic similarity, reasoning, and query expansion, leading to improved interpretability and relevance.
claimTraditional question-answering systems match textual questions with answers in unstructured text databases by analyzing semantic relationships and matching questions to answers with maximum semantic similarity.
Medical Hallucination in Foundation Models and Their Impact on ... medrxiv.org Nov 2, 2025 4 facts
measurementThe advanced reasoning model gpt-5 achieves a 71.2% baseline resistance to hallucinations and a semantic similarity score greater than 0.8.
claimMedical-specialized models may be memorizing surface-level medical terminology without developing the deeper relational understanding necessary for reliable reasoning about complex clinical scenarios, as evidenced by their failure to achieve high semantic similarity scores despite explicit medical pretraining.
claimHallucination resistance in AI models correlates more strongly with the depth of conceptual understanding, as measured by semantic similarity to ground truth, than with exposure to domain-specific training data.
measurementThe models gemini-2.5-pro, o3-mini, and deepseek-r1 cluster in the high semantic similarity range of 0.8–0.9, indicating strong semantic alignment with ground truth medical information.
A Knowledge Graph-Based Hallucination Benchmark for Evaluating ... arxiv.org Feb 23, 2026 3 facts
procedureThe entity-level filter classifies responses as aligned, hallucinated, or abstained by identifying abstentions and evaluating semantic and token-level similarities against an entity's description.
procedureThe entity-level filter evaluates semantic similarity using cosine similarity on encoded representations of the response and the entity description, and evaluates token-level similarity using the intersection of common words.
measurementThe entity-level filter combines semantic and token-level similarity metrics using a 70:30 ratio, prioritizing semantic over lexical alignment.
Knowledge Graphs vs RAG: When to Use Each for AI in 2026 - Atlan atlan.com Feb 12, 2026 2 facts
claimVector-based RAG systems use semantic similarity to retrieve information, which can lead to retrieving text that mentions entities together without confirming if they have a functional relationship, such as being purchased together.
claimKnowledge graphs utilize graph traversal following explicit relationships for retrieval, while RAG systems utilize semantic similarity search across vector space.
Detect hallucinations for RAG-based systems - AWS aws.amazon.com May 16, 2025 2 facts
claimFor use cases where precision is the highest priority, the token similarity, LLM prompt-based, and semantic similarity methods are recommended, whereas the BERT stochastic method outperforms other methods for high recall.
claimSemantic similarity and token similarity detectors for hallucination detection show very low accuracy and recall but perform well with regards to precision, indicating they are primarily useful for identifying the most evident hallucinations.
Construction of Knowledge Graphs: State and Challenges - arXiv arxiv.org 2 facts
claimPre-trained word embeddings assist in the Entity Linking disambiguation process by encoding semantic similarity within a latent space.
claimDictionaries and pre-trained word embeddings are used in ontology matching to capture semantic similarity between words that are not character-level similar.
Medical Hallucination in Foundation Models and Their ... medrxiv.org Mar 3, 2025 2 facts
claimThere is a robust correlation between semantic similarity and hallucination resistance in LLMs, suggesting that a deeper understanding of medical concepts is a critical factor in minimizing factual errors in generated medical content.
procedureContextual Relevance Evaluation addresses Context Deviation Issues by using n-gram overlap and semantic similarity metrics to assess whether Large Language Model outputs maintain appropriate clinical context.
How to Improve Multi-Hop Reasoning With Knowledge Graphs and ... neo4j.com Jun 18, 2025 1 fact
claimGraphRAG combines semantic similarity via vector search with structured reasoning via graph queries to enable LLMs to deliver relevant, traceable answers.
Construction of intelligent decision support systems through ... - Nature nature.com Oct 10, 2025 1 fact
claimMulti-criteria alignment scoring uses semantic, structural, and functional similarity measures to create mappings between domain concepts.
Combining Knowledge Graphs With LLMs | Complete Guide - Atlan atlan.com Jan 28, 2026 1 fact
claimVector embeddings capture semantic similarity between data points but fail to capture explicit relationships between entities, whereas knowledge graphs provide structured connections that vector search cannot infer.
The construction and refined extraction techniques of knowledge ... nature.com Feb 10, 2026 1 fact
measurementThe knowledge graph showed an average semantic similarity of 0.92 to expert-annotated references when evaluated via BERTScore on a subset of 10,000 triplets.