Arabic Syntactic Trees: from Constituency to Dependency

What it is about

The note reports work in progress on converting Penn-style phrase-structure trees of Arabic (from the LDC Arabic Treebank) into the analytical dependency trees of the Prague school. The conversion has two parts: a recursive function that turns the topology of a phrase tree into a dependency tree by picking a head child at each node, and a step that assigns analytical functions to the nodes from their part of speech, their ancestors' labels and their parent. Traces are resolved by moving the co-indexed subtree into the place of the trace node. The conversion is worked through one model sentence, and the preliminary evaluation reports about 60 percent accuracy for the tree topology and about the same for function assignment. The authors expect 85–90 percent once more phenomena are handled.

Ideas

• I1. A head-percolation table for Arabic phrase trees as data, not code: each phrase label lists its head child, and the converter is the recursive function described in the note. Build the same table for treebanks/ar from the Arabic phrase structure of a corpus, and test it on the model sentence.
• I2. Trace resolution as a rule: a clause co-indexed with an empty subject trace is pruned from its position and attached where the trace stood. Test it on the trace examples the paper gives (NP-SBJ and NP-SBJ-1 with a moved topic).
• I3. State clauses after a nominal predicate: the head-first algorithm wrongly makes a state clause a sister of the main predicate, and the paper needs a corrective reshuffle. Write the reshuffle as an explicit rule (a clause whose predicate is non-verbal is attached under the nominal it describes) and test it on the model sentence's "huwa yas'adu" clause and on similar clauses.

For Mova

I1 and I2 apply to treebanks/ar (the conversion step, if an Arabic treebank is built from phrase structure). I3 is a rule for the nominal-clause cases in 7. Syntax and Syntax: the clause and the sentence. The paper's figures and sentence are NC-licensed summary material: use the idea, not the figures.

Topics: Treebanks

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@inproceedings{zabokrtsky2003arabic,
  title = {{Arabic Syntactic Trees: from Constituency to Dependency}},
  author = {Zabokrtsky, Zdenek and Smrz, Otakar},
  booktitle = {10th Conference of the European Chapter of the Association for Computational Linguistics},
  year = {2003},
  url = {https://aclanthology.org/E03-1014/}
}

References

8 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Haim Gaifman. 1965. Dependency Systems and Phrase-Structure Systems. Information and Control, pages 304-337. find
2. Jan Haile, Eva Hajieova, Petr Pajas, Jarmila Panevova, Petr Sgall, and Barbora Vidova-Hladka. 2001. Prague Dependency Treebank 1.0 (Final Production Label). CDROM CAT: LDC2001T10, ISBN 1-58563-212-0. find
3. Mohamed Maamouri and Christopher Cieri. 2002. Resources for Natural Language Processing at the Linguistic Data Consortium. In Proceedings of the International Symposium on Processing of Arabic, pages 125-146, Tunisia, April 18th-20th. Faculte des Lettres, University of Manouba. find
4. Mohamed Maamouri, Ann Bies, Hubert Jin, and Tim Buckwalter. 2003. Arabic Treebank: Part 1 v 2.0. LDC catalog number LDC2003T06, ISBN 1-58563-261-9. find
5. Gerold Schneider. 1998. A Linguistic Comparison Constituency, Dependency, and Link Grammar. Master's thesis, University of Zurich. find
6. Otakar Smil, Jan 'litaidauf, and Petr Zemanek. 2002. Prague Dependency Treebank for Arabic: Multi-Level Annotation of Arabic Corpus. In Proceedings of the International Symposium on Processing of Arabic, pages 147-155, Tunisia, April 18th-20th. Faculte des Lettres, University of Manouba. find
7. Fei Xia and Martha Palmer. 2001. Converting Dependency Structures to Phrase Structures. In Proceedings of the Human Language Technology Conference (HLT-2001), San Diego, CA, March 18-21. find
8. Zdetlek abokrtsk3 and Ivona Kue'erova. 2002. Transforming Penn Treebank Phrase Trees into (Praguian) Tectogrammatical Dependency Trees. Prague Bulletin of Mathematical Linguistics, (78):77-94. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/map-2003-zabokrtsky-arabic-syntactic-trees-constituency-dependency/