Learning PP attachment for filtering prosodic phrasing

O. van Herwijnen, A. van den Bosch, J.M.B. Terken, E.C. Marsi

    Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

    Abstract

    We explore learning prepositional-phrase attachment in Dutch, to use it as a filter in prosodic phrasing. From a syntactic treebank of spoken Dutch we extract instances of the attachment of prepositional phrases to either a governing verb or noun. Using cross-validated parameter and feature selection, we train two learning algorithms, IB1 and Ripper, on making this distinction, based on unigram and bigram lexical features and a cooccurrence feature derived from WWW counts. We optimize the learning on noun attachment, since in a second stage we use the attachment decision for blocking the incorrect placement of phrase boundaries before prepositional phrases attached to the preceding noun. On noun attachment, IB1 attains an F-score of 83; Ripper an F-score of 79. When used as a filter for prosodic phrasing, using attachment decisions from IB1 yields the best improvement on recall (by four points to 69 F-score) on phrase boundary placement.
    Original languageEnglish
    Title of host publicationProceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary
    Place of PublicationNew Brunswick, NJ
    PublisherACL
    Pages139-146
    ISBN (Print)1932432000
    Publication statusPublished - 2003

    Fingerprint

    Syntactics
    World Wide Web
    Learning algorithms
    Feature extraction

    Cite this

    van Herwijnen, O., van den Bosch, A., Terken, J. M. B., & Marsi, E. C. (2003). Learning PP attachment for filtering prosodic phrasing. In Proceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary (pp. 139-146). New Brunswick, NJ: ACL.
    van Herwijnen, O. ; van den Bosch, A. ; Terken, J.M.B. ; Marsi, E.C. / Learning PP attachment for filtering prosodic phrasing. Proceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary. New Brunswick, NJ : ACL, 2003. pp. 139-146
    @inproceedings{8486ffcd74f646baa2cc352747e3f485,
    title = "Learning PP attachment for filtering prosodic phrasing",
    abstract = "We explore learning prepositional-phrase attachment in Dutch, to use it as a filter in prosodic phrasing. From a syntactic treebank of spoken Dutch we extract instances of the attachment of prepositional phrases to either a governing verb or noun. Using cross-validated parameter and feature selection, we train two learning algorithms, IB1 and Ripper, on making this distinction, based on unigram and bigram lexical features and a cooccurrence feature derived from WWW counts. We optimize the learning on noun attachment, since in a second stage we use the attachment decision for blocking the incorrect placement of phrase boundaries before prepositional phrases attached to the preceding noun. On noun attachment, IB1 attains an F-score of 83; Ripper an F-score of 79. When used as a filter for prosodic phrasing, using attachment decisions from IB1 yields the best improvement on recall (by four points to 69 F-score) on phrase boundary placement.",
    author = "{van Herwijnen}, O. and {van den Bosch}, A. and J.M.B. Terken and E.C. Marsi",
    year = "2003",
    language = "English",
    isbn = "1932432000",
    pages = "139--146",
    booktitle = "Proceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary",
    publisher = "ACL",

    }

    van Herwijnen, O, van den Bosch, A, Terken, JMB & Marsi, EC 2003, Learning PP attachment for filtering prosodic phrasing. in Proceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary. ACL, New Brunswick, NJ, pp. 139-146.

    Learning PP attachment for filtering prosodic phrasing. / van Herwijnen, O.; van den Bosch, A.; Terken, J.M.B.; Marsi, E.C.

    Proceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary. New Brunswick, NJ : ACL, 2003. p. 139-146.

    Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

    TY - GEN

    T1 - Learning PP attachment for filtering prosodic phrasing

    AU - van Herwijnen, O.

    AU - van den Bosch, A.

    AU - Terken, J.M.B.

    AU - Marsi, E.C.

    PY - 2003

    Y1 - 2003

    N2 - We explore learning prepositional-phrase attachment in Dutch, to use it as a filter in prosodic phrasing. From a syntactic treebank of spoken Dutch we extract instances of the attachment of prepositional phrases to either a governing verb or noun. Using cross-validated parameter and feature selection, we train two learning algorithms, IB1 and Ripper, on making this distinction, based on unigram and bigram lexical features and a cooccurrence feature derived from WWW counts. We optimize the learning on noun attachment, since in a second stage we use the attachment decision for blocking the incorrect placement of phrase boundaries before prepositional phrases attached to the preceding noun. On noun attachment, IB1 attains an F-score of 83; Ripper an F-score of 79. When used as a filter for prosodic phrasing, using attachment decisions from IB1 yields the best improvement on recall (by four points to 69 F-score) on phrase boundary placement.

    AB - We explore learning prepositional-phrase attachment in Dutch, to use it as a filter in prosodic phrasing. From a syntactic treebank of spoken Dutch we extract instances of the attachment of prepositional phrases to either a governing verb or noun. Using cross-validated parameter and feature selection, we train two learning algorithms, IB1 and Ripper, on making this distinction, based on unigram and bigram lexical features and a cooccurrence feature derived from WWW counts. We optimize the learning on noun attachment, since in a second stage we use the attachment decision for blocking the incorrect placement of phrase boundaries before prepositional phrases attached to the preceding noun. On noun attachment, IB1 attains an F-score of 83; Ripper an F-score of 79. When used as a filter for prosodic phrasing, using attachment decisions from IB1 yields the best improvement on recall (by four points to 69 F-score) on phrase boundary placement.

    M3 - Conference contribution

    SN - 1932432000

    SP - 139

    EP - 146

    BT - Proceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary

    PB - ACL

    CY - New Brunswick, NJ

    ER -

    van Herwijnen O, van den Bosch A, Terken JMB, Marsi EC. Learning PP attachment for filtering prosodic phrasing. In Proceedings of the 10th Conference of the European Chapter of the Association for Computational Linguistics, april 12-17, 2003, Budapest, Hungary. New Brunswick, NJ: ACL. 2003. p. 139-146