Syntax Meets Semantics: Understanding Scientific Formulae
2026-08-03 • Information Retrieval
Information RetrievalArtificial Intelligence
AI summaryⓘ
The authors studied how scientific formulas can be understood both by their structure (syntax) and their meaning (semantics). They found that the usual ways of representing these two aspects don't seem closely connected at first, even though there is an underlying link. By using special methods that learn from both forms together, they improved how well computers can match formula structure with meaning. This shows that new learning techniques can help bridge the gap between formula syntax and semantics for better information search.
scientific formulaesyntaxsemanticsinformation retrievalrepresentation learningcontrastive learninggraph-based encoderstext-based encoderscross-modal retrieval
Authors
Yuni Susanti, Moritz Schubotz
Abstract
Scientific formulae are a fundamental component of scholarly communication, yet their dual nature -- as structured syntax and carriers of semantics -- remains underexplored in scholarly information retrieval. Although prior studies show that jointly modeling syntactic and semantic modalities improves retrieval performance, the relationship between their underlying representations has not been systematically investigated. In this work, we empirically study cross-modal correspondence between formula syntax and semantics. We find that their native representation spaces exhibit extremely weak observable correspondence despite strong latent correlation, indicating a substantial representation mismatch between the two modalities. We further evaluate whether this mismatch can be reduced using standard representation learning and alignment techniques. We represent syntactic structure using graph-based encoders and semantic information using text-based encoders, then apply contrastive learning to induce a shared representation space. Results show that the learned alignment substantially improves cross-modal retrieval, suggesting that explicit representation learning can recover correspondence absent from the original representation spaces.