--- title: "Semantic Similarity | Intro to Semantics and Pragmatics" description: "Semantic similarity is the degree to which meanings overlap across words, phrases, or sentences, and it’s central to corpus-based analysis in semantics." canonical: "https://fiveable.me/introduction-semantics-pragmatics/key-terms/semantic-similarity" type: "key-term" subject: "Intro to Semantics and Pragmatics" unit: "Unit 15" --- # Semantic Similarity | Intro to Semantics and Pragmatics ## Definition Semantic similarity is how close two words, phrases, or sentences are in meaning. In Intro to Semantics and Pragmatics, it shows up in corpus-based and computational semantics when you compare meanings using context and language data. ## What It Is Semantic similarity is the degree of shared meaning between linguistic expressions in Intro to Semantics and Pragmatics. You use it when two words, phrases, or sentences are not identical in meaning, but still point to related ideas, like how "doctor" and "physician" are very close, while "doctor" and "nurse" are related but less similar. In this course, the term matters because meaning is not treated as just a dictionary entry. Semantic similarity depends on how expressions pattern in real language use, so it connects to corpus-based semantics, where researchers look at large collections of text and ask what kinds of contexts words appear in. If two expressions show up in similar environments, a model may treat them as semantically similar. That is why semantic similarity is often discussed alongside vector representations of meaning. Word embeddings, for example, place words in a multi-dimensional space, and words with closer vectors are usually treated as more similar. A computer does not "know" meaning the way a person does, but it can estimate similarity from patterns of co-occurrence and context. A simple classroom example is comparing "buy a car" with "purchase a vehicle." These are not the same string of words, but they are close in meaning, so a semantic similarity model should score them higher than "buy a car" and "read a poem." That makes the idea useful for seeing how computational systems handle paraphrase, synonymy, and relatedness. One thing to watch is that semantic similarity is not always the same as simple word overlap. Two sentences can share few words and still be very similar in meaning, while two sentences can share many words and mean very different things. In this course, that difference is part of the bigger question of how language meaning is represented beyond surface form. ## Why It Matters Semantic similarity gives you a way to talk about meaning with evidence instead of just intuition. In Intro to Semantics and Pragmatics, it bridges theory and data, since you can compare how people think words relate and how corpus-based models measure those relationships. It matters for understanding why computational semantics works the way it does. When a system searches for related content, translates text, or summarizes a passage, it has to estimate which expressions are near each other in meaning. Semantic similarity is the idea that lets those systems rank one phrase as a better match than another. It also sharpens your reading of semantic claims in class. If a question asks whether two expressions are synonymous, merely associated, or only contextually close, semantic similarity helps you separate those possibilities. That distinction shows up whenever you compare dictionary meaning, usage in context, and model output from a corpus. For assignments, this term often appears in examples involving word vectors, text comparison, or simple corpus analysis. If you can explain why two items are similar, and what kind of similarity they share, you are already doing the kind of analysis this part of the course expects. ## Connections ### Word Embeddings Word embeddings are one way to represent semantic similarity mathematically. They place words in a vector space so that closer points usually correspond to closer meanings. In this course, embeddings are the bridge between raw text and a model that can compare meanings across lots of examples. ### [Cosine Similarity](/introduction-semantics-pragmatics/key-terms/cosine-similarity) Cosine similarity is a common method for measuring how close two vectors are, which makes it a standard tool for estimating semantic similarity in computational semantics. Instead of comparing word definitions directly, it compares the angle between vectors. That makes it useful when you want a numerical score for meaning overlap. ### [annotated corpus](/introduction-semantics-pragmatics/key-terms/annotated-corpus) An annotated corpus gives you text with labels that can support semantic analysis, like tagged meanings, parts of speech, or sense distinctions. Semantic similarity often gets tested on corpus data because you can look at patterns of use instead of guessing from isolated examples. The annotations help make those comparisons more precise. ### [Natural Language Processing](/introduction-semantics-pragmatics/key-terms/natural-language-processing) Natural Language Processing uses semantic similarity in tasks like search, translation, and summarization. If an NLP system can tell that two phrases mean nearly the same thing, it can match queries better and produce cleaner outputs. In this course, NLP is where semantic theory becomes a working computational method. ## On the AP Exam A quiz question might give you two sentences, a pair of words, or a small corpus example and ask which items are most semantically similar. You would compare meaning, not just shared vocabulary, and explain whether the relationship is synonymy, close association, or broader topical relatedness. If the prompt includes a model or vector output, you may need to interpret which pair has the higher similarity score and why. In short-answer or discussion responses, this term shows up when you explain how corpus evidence supports a meaning judgment. A strong answer names the context pattern, the similarity method, or the model behavior, then links that back to the linguistic interpretation. ## semantic similarity vs semantic relatedness Semantic similarity is narrower than semantic relatedness. Similarity means two expressions are close in meaning, like synonym-like pairs, while relatedness can include looser connections such as part-whole links or common topics. For example, "car" and "automobile" are similar, but "car" and "road" are related without being very similar. ## Key Takeaways - Semantic similarity measures how close two expressions are in meaning, not just whether they share the same words. - In Intro to Semantics and Pragmatics, the term connects directly to corpus-based semantics and computational models of meaning. - Word embeddings and cosine similarity are common ways to estimate semantic similarity from large text collections. - Two items can be semantically similar even if they do not look alike on the surface, which is why context matters. - The term is useful when you need to compare meanings, rank search results, or explain how language models group words together. ## FAQs ### What is semantic similarity in Intro to Semantics and Pragmatics? Semantic similarity is the degree to which two words, phrases, or sentences share meaning. In this course, it matters because semanticists and computational linguists use it to compare expressions using context, corpora, and vector-based models. ### How is semantic similarity different from semantic relatedness? Semantic similarity is about closeness in meaning, while semantic relatedness is broader and can include topical or associative links. "Doctor" and "physician" are similar, but "doctor" and "hospital" are related without being very similar. ### How do you measure semantic similarity? A common approach is to represent words or phrases as vectors and then compare them with cosine similarity. Other methods use overlap in context or corpus patterns. The main idea is to turn meaning into something you can compare numerically. ### Why do word embeddings matter for semantic similarity? Word embeddings model meaning by placing words in a space where nearby vectors tend to have related meanings. That lets a system estimate similarity from usage patterns in text, which is why embeddings show up in computational semantics and NLP. ## Related Study Guides - [15.3 Corpus-based and computational semantics](/introduction-semantics-pragmatics/unit-15/corpus-based-computational-semantics/study-guide/xomd8PCVk67SwS8b) ## About This Document Canonical Fiveable pages are available as Markdown at the same path plus `.md`. - [llms.txt](https://fiveable.me/llms.txt): index of Fiveable's sections and URL patterns - [llms-full.txt](https://fiveable.me/llms-full.txt): complete subject and unit listing - [MCP server](https://fiveable.me/mcp): call Fiveable as tools instead of fetching pages (`https://fiveable.me/api/mcp`) - [MCP server for AP teachers](https://fiveable.me/mcp/teachers): a teacher's classes, assignments and AP-rubric grading (`https://fiveable.me/api/mcp/teacher`) ## Structured Data ```json {"@context":"https://schema.org","@graph":[{"@type":"LearningResource","@id":"https://fiveable.me/introduction-semantics-pragmatics/key-terms/semantic-similarity#resource","name":"Semantic Similarity | Intro to Semantics and Pragmatics","url":"https://fiveable.me/introduction-semantics-pragmatics/key-terms/semantic-similarity","learningResourceType":"Concept explainer","educationalLevel":"AP® / High School","about":{"@id":"https://fiveable.me/introduction-semantics-pragmatics/key-terms/semantic-similarity#term"},"audience":{"@type":"EducationalAudience","educationalRole":"student"},"dateModified":"2026-07-03T02:22:58.641Z","isPartOf":{"@type":"Collection","name":"Intro to Semantics and Pragmatics Key Terms","url":"https://fiveable.me/introduction-semantics-pragmatics/key-terms"},"publisher":{"@type":"Organization","name":"Fiveable","url":"https://fiveable.me"}},{"@type":"DefinedTerm","@id":"https://fiveable.me/introduction-semantics-pragmatics/key-terms/semantic-similarity#term","name":"semantic similarity","description":"Semantic similarity is how close two words, phrases, or sentences are in meaning. In Intro to Semantics and Pragmatics, it shows up in corpus-based and computational semantics when you compare meanings using context and language data.","url":"https://fiveable.me/introduction-semantics-pragmatics/key-terms/semantic-similarity","inDefinedTermSet":{"@type":"DefinedTermSet","name":"Intro to Semantics and Pragmatics Key Terms","url":"https://fiveable.me/introduction-semantics-pragmatics/key-terms"}},{"@type":"FAQPage","mainEntity":[{"@type":"Question","name":"What is semantic similarity in Intro to Semantics and Pragmatics?","acceptedAnswer":{"@type":"Answer","text":"Semantic similarity is the degree to which two words, phrases, or sentences share meaning. In this course, it matters because semanticists and computational linguists use it to compare expressions using context, corpora, and vector-based models."}},{"@type":"Question","name":"How is semantic similarity different from semantic relatedness?","acceptedAnswer":{"@type":"Answer","text":"Semantic similarity is about closeness in meaning, while semantic relatedness is broader and can include topical or associative links. \"Doctor\" and \"physician\" are similar, but \"doctor\" and \"hospital\" are related without being very similar."}},{"@type":"Question","name":"How do you measure semantic similarity?","acceptedAnswer":{"@type":"Answer","text":"A common approach is to represent words or phrases as vectors and then compare them with cosine similarity. Other methods use overlap in context or corpus patterns. The main idea is to turn meaning into something you can compare numerically."}},{"@type":"Question","name":"Why do word embeddings matter for semantic similarity?","acceptedAnswer":{"@type":"Answer","text":"Word embeddings model meaning by placing words in a space where nearby vectors tend to have related meanings. That lets a system estimate similarity from usage patterns in text, which is why embeddings show up in computational semantics and NLP."}}]},{"@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Intro to Semantics and Pragmatics","item":"https://fiveable.me/introduction-semantics-pragmatics"},{"@type":"ListItem","position":2,"name":"Key Terms","item":"https://fiveable.me/introduction-semantics-pragmatics/key-terms"},{"@type":"ListItem","position":3,"name":"Unit 15","item":"https://fiveable.me/introduction-semantics-pragmatics/unit-15"},{"@type":"ListItem","position":4,"name":"semantic similarity"}]}]} ```