Elamur Research⌕
☰
home / papers / 2026 / Fast and Flexible Example-Based Treebank Search with Vector Symbolic Architectures

Fast and Flexible Example-Based Treebank Search with Vector Symbolic Architectures

Paper · 2026 · Adam Roussel

Summary

Querying dependency treebanks usually relies on dedicated query languages in which every constraint must hold. This makes it hard to explore constructions whose exact form is unclear or naturally varies. Roussel proposes an example-based search method built on vector symbolic architectures (VSA), also called hyperdimensional computing. Each sentence's parse tree is turned into one high-dimensional binary vector. Tokens are bundles of UD annotation fields, and dependency relations are encoded as permuted key paths. A query example is encoded the same way. The system then ranks every treebank sentence by vector similarity, so all combinations of query features are effectively tested at once. Matching gives graded scores rather than a strict yes/no decision. Mismatching parts act only as noise, which also helps when sentences have been parsed incorrectly. Every query costs one vector comparison per sentence, so runtime grows linearly with treebank size and does not depend on query complexity. The tool is lightweight enough for ordinary local hardware. The author presents example outputs and argues that the design could later include semantic information such as word embeddings.

Key points

  • The implementation uses binary spatter code (BSC) with dense binary vectors. Binding is XOR, bundling is element-wise majority, and similarity is based on Hamming distance.
  • Token bundles combine the UD fields form, lemma, upos, xpos and any feats. Key vectors are permuted so they stay distinct from values, which avoids self-cancellation under XOR.
  • A "Subtree" encoding places each token's features at every dependency path leading to it. This lets properties match regardless of their depth in the tree, at the cost of larger bundles.
  • The method is positioned against constraint-based tools (ICARUS, SETS) and example-based ones (Linguist's Search Engine, GrETEL). In those example-based tools, users must loosen constraints by hand.
  • The code is released as vsatreebank on Codeberg under GPLv3. Planned future work includes embedding features and moving to other VSA types. The research was funded by DFG SFB 1475.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Ruhr University BochumBochum · Germany25 works · 1,189 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 1Research subjects and languages the work is about.Treebanks · 126 works

Cite

Cite as Adam Roussel. 2026. Fast and Flexible Example-Based Treebank Search with Vector Symbolic Architectures. Proceedings of the Workshop on Structured Linguistic Data and Evaluation (SLiDE).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/26p8zoerpa3d

@inproceedings{roussel2026fast,
  title = {{Fast and Flexible Example-Based Treebank Search with Vector Symbolic Architectures}},
  author = {Adam Roussel},
  booktitle = {Proceedings of the Workshop on Structured Linguistic Data and Evaluation (SLiDE)},
  year = {2026},
  doi = {10.63317/26p8zoerpa3d},
  url = {https://doi.org/10.63317/26p8zoerpa3d}
}

References

13 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Denis Kleyko, Dmitri A. Rachkovskij, Evgeny Osipov, and Abbas Rahimi. 2022. A survey on hyperdimensional computing aka. vector symbolic architectures, part I: Models and data transformations. ACM Comput. Surv., 55(6):1–40. find
  2. Liesbeth Augustinus, Vincent Vandeghinste, and Frank Van Eynde. 2012. Example-based treebank querying. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC’12), pages 3161–3167, Istanbul, Turkey. European Language Resources Association (ELRA). find
  3. Juhani Luotolahti, Jenna Kanerva, Sampo Pyysalo, and Filip Ginter. 2015. SETS: Scalable and efficient tree search in dependency graphs. In Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Demonstrations, pages 51–55, Denver, Colorado.… find
  4. Trevor Cohen, Dominic Widdows, Manuel Wahle, and Roger Schvaneveldt. 2014. Orthogonality and Orthography: Introducing Measured Distance into Semantic Space, pages 34–46. Springer Berlin Heidelberg. find
  5. Jan Odijk, Martijn van der Klis, and Sheean Spoel. 2017. Extensions to the GrETEL treebank query application. In Proceedings of the 16th International Workshop on Treebanks and Linguistic Theories, pages 46–55, Prague, Czech Republic. find
  6. Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre, and Daniel Zeman. 2021. Universal Dependencies. Computational Linguistics, pages 1–54.
  7. Philip Resnik and Aaron Elkiss. 2005. The Linguist’s Search Engine: An overview. In Proceedings of the ACL Interactive Poster and Demonstration Sessions, pages 33–36, Ann Arbor, Michigan. Association for Computational Linguistics. find
  8. Chris Eliasmith. 2013. How to Build a Brain: A Neural Architecture for Biological Cognition. Oxford University Press. find
  9. Markus Gärtner, Gregor Thiele, Wolfgang Seeker, Anders Björkelund, and Jonas Kuhn. 2013. find
  10. Kenny Schlegel, Peer Neubert, and Peter Protzel. 2022. A comparison of vector symbolic architectures. 55:4523–4555. find
  11. Hans-Jörg Schmid. 2000. English Abstract Nouns as Conceptual Shells: From Corpus to Cognition. De Gruyter. find
  12. Gertjan van Noord, Gosse Bouma, Frank Van Eynde, Daniël de Kok, Jelmer van der Linde, Ineke Schuurman, Erik Tjong Kim Sang, and Vincent Vandeghinste. 2012. Large Scale Syntactic Annotation of Written Dutch: Lassy, page 147–164. Springer Berlin Heidelberg. find
  13. D. Widdows and T. Cohen. 2015. Reasoning with vectors: A continuous model for fast robust inference. 23(2):141–173. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/roussel-2026-fast-flexible-example-based-treebank/