# Beyond Cosine Similarity: Taming Semantic Drift and Antonym Intrusion in a 15-Million Node Turkish Synonym Graph

arXiv: https://arxiv.org/abs/2601.13251 · HTML: https://mlsanity.com/paper/2601.13251/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 44 references that are in the map.
- First posted on arXiv 19 Jan 2026.

## Builds on (most cited first, 8 of 8)

- [Billion-scale similarity search with GPUs](https://mlsanity.com/paper/1702.08734/) (2017)
- [Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere](https://mlsanity.com/paper/2005.10242/) (2020)
- [From Louvain to Leiden: guaranteeing well-connected communities](https://mlsanity.com/paper/1810.08473/) (2018)
- [Counter-fitting Word Vectors to Linguistic Constraints](https://mlsanity.com/paper/1603.00892/) (2016)
- [Exploring the Representation of Word Meanings in Context: A Case Study on Homonymy and Synonymy](https://mlsanity.com/paper/2106.13553/) (2021)
- [Computing Lexical Contrast](https://mlsanity.com/paper/1308.6300/) (2013)
- [Watset: Automatic Induction of Synsets from a Graph of Synonyms](https://mlsanity.com/paper/1704.07157/) (2017)
- [TurkEmbed4Retrieval: Turkish Embedding Model for Retrieval Task](https://mlsanity.com/paper/2511.07595/) (2025)

## Abstract (arXiv)

Neural embeddings have a notorious blind spot: they can't reliably tell synonyms apart from antonyms. Consequently, increasing similarity thresholds often fails to prevent opposites from being grouped together. We've built a large-scale semantic clustering system specifically designed to tackle this problem head on. Our pipeline chews through 15 million lexical items, evaluates a massive 520 million potential relationships, and ultimately generates 2.9 million high-precision semantic clusters. The system makes three primary contributions. First, we introduce a labeled dataset of 843,000 concept pairs spanning synonymy, antonymy, and co-hyponymy, constructed via Gemini 2.5-Flash LLM augmentation and verified using human-curated dictionary resources. Second, we propose a specialized three-way semantic relation discriminator that achieves 90% macro-F1, enabling robust disambiguation beyond raw embedding similarity. Third, we introduce a novel soft-to-hard clustering algorithm that mitigates semantic drift preventing erroneous transitive chains (e.g., hot -> spicy -> pain -> depression) while simultaneously resolving polysemy. Our approach employs a topology-aware two-stage expansion-pruning procedure with topological voting, ensuring that each term is assigned to exactly one semantically coherent cluster. The resulting resource enables high-precision semantic search and retrieval-augmented generation, particularly for morphologically rich and low-resource languages where existing synonym databases remain sparse.