Interactive Translation Prediction versus Conventional Post-editing in Practice: A Study with the CasMaCat Workbench

German Sanchis-Trilles, Vicent Alabau, Christian Buck, Michael Carl, Francisco Casacuberta, Mercedes Garcia Martinez, Ulrich Germann, Jesus Gonzalez-Rubio, Robin L. Hill, Philipp Koehn, Luis A. Leiva, Bartolomé Mesa-Lao, Daniel Ortiz-Martnez, Herve Saint-Amand, Chara Tsoukala

    Research output: Contribution to journalJournal articleResearchpeer-review

    Abstract

    We conducted a field trial in computer-assisted professional translation to compare Interactive Translation Prediction (ITP) against conventional post-
    editing (PE) of machine translation (MT) output. In contrast to the conventional
    PE set-up, where an MT system first produces a static translation hypothesis that is then edited by a professional translator (hence \post-editing"), ITP constantly updates the translation hypothesis in real time in response to user edits. Our study involved nine professional translators and four reviewers working with the web-based CasMaCat workbench. Various new interactive features aiming to assist the post-editor were also tested in this trial. Our results show that even with little training, ITP can be as productive as conventional PE in terms of the total time required to produce the final translation. Moreover, in the ITP setting translators require fewer key strokes to arrive at the final version of their translation.
    Original languageEnglish
    JournalMachine Translation
    Volume28
    Issue number3-4
    Pages (from-to)217-235
    ISSN0922-6567
    DOIs
    Publication statusPublished - 2014

    Keywords

    • CAT
    • SMT
    • Interactive translation prediction
    • Post-editing
    • Field trial
    • User studies

    Cite this

    Sanchis-Trilles, G., Alabau, V., Buck, C., Carl, M., Casacuberta, F., Martinez, M. G., ... Tsoukala, C. (2014). Interactive Translation Prediction versus Conventional Post-editing in Practice: A Study with the CasMaCat Workbench. Machine Translation, 28(3-4), 217-235. https://doi.org/10.1007/s10590-014-9157-9
    Sanchis-Trilles, German ; Alabau, Vicent ; Buck, Christian ; Carl, Michael ; Casacuberta, Francisco ; Martinez, Mercedes Garcia ; Germann, Ulrich ; Gonzalez-Rubio, Jesus ; Hill, Robin L. ; Koehn, Philipp ; A. Leiva, Luis ; Mesa-Lao, Bartolomé ; Ortiz-Martnez, Daniel ; Saint-Amand, Herve ; Tsoukala, Chara. / Interactive Translation Prediction versus Conventional Post-editing in Practice : A Study with the CasMaCat Workbench. In: Machine Translation. 2014 ; Vol. 28, No. 3-4. pp. 217-235.
    @article{a07b37ef0bdd4ac396eefdfbb156bc75,
    title = "Interactive Translation Prediction versus Conventional Post-editing in Practice: A Study with the CasMaCat Workbench",
    abstract = "We conducted a field trial in computer-assisted professional translation to compare Interactive Translation Prediction (ITP) against conventional post-editing (PE) of machine translation (MT) output. In contrast to the conventionalPE set-up, where an MT system first produces a static translation hypothesis that is then edited by a professional translator (hence \post-editing{"}), ITP constantly updates the translation hypothesis in real time in response to user edits. Our study involved nine professional translators and four reviewers working with the web-based CasMaCat workbench. Various new interactive features aiming to assist the post-editor were also tested in this trial. Our results show that even with little training, ITP can be as productive as conventional PE in terms of the total time required to produce the final translation. Moreover, in the ITP setting translators require fewer key strokes to arrive at the final version of their translation.",
    keywords = "CAT, SMT, Interactive translation prediction, Post-editing, Field trial, User studies, CAT, SMT, Interactive translation prediction, Post-editing, Field trial, User studies",
    author = "German Sanchis-Trilles and Vicent Alabau and Christian Buck and Michael Carl and Francisco Casacuberta and Martinez, {Mercedes Garcia} and Ulrich Germann and Jesus Gonzalez-Rubio and Hill, {Robin L.} and Philipp Koehn and {A. Leiva}, Luis and Bartolom{\'e} Mesa-Lao and Daniel Ortiz-Martnez and Herve Saint-Amand and Chara Tsoukala",
    year = "2014",
    doi = "10.1007/s10590-014-9157-9",
    language = "English",
    volume = "28",
    pages = "217--235",
    journal = "Machine Translation",
    issn = "0922-6567",
    publisher = "Springer",
    number = "3-4",

    }

    Sanchis-Trilles, G, Alabau, V, Buck, C, Carl, M, Casacuberta, F, Martinez, MG, Germann, U, Gonzalez-Rubio, J, Hill, RL, Koehn, P, A. Leiva, L, Mesa-Lao, B, Ortiz-Martnez, D, Saint-Amand, H & Tsoukala, C 2014, 'Interactive Translation Prediction versus Conventional Post-editing in Practice: A Study with the CasMaCat Workbench', Machine Translation, vol. 28, no. 3-4, pp. 217-235. https://doi.org/10.1007/s10590-014-9157-9

    Interactive Translation Prediction versus Conventional Post-editing in Practice : A Study with the CasMaCat Workbench. / Sanchis-Trilles, German; Alabau, Vicent; Buck, Christian; Carl, Michael; Casacuberta, Francisco; Martinez, Mercedes Garcia; Germann, Ulrich; Gonzalez-Rubio, Jesus; Hill, Robin L.; Koehn, Philipp; A. Leiva, Luis; Mesa-Lao, Bartolomé ; Ortiz-Martnez, Daniel; Saint-Amand, Herve; Tsoukala, Chara.

    In: Machine Translation, Vol. 28, No. 3-4, 2014, p. 217-235.

    Research output: Contribution to journalJournal articleResearchpeer-review

    TY - JOUR

    T1 - Interactive Translation Prediction versus Conventional Post-editing in Practice

    T2 - A Study with the CasMaCat Workbench

    AU - Sanchis-Trilles, German

    AU - Alabau, Vicent

    AU - Buck, Christian

    AU - Carl, Michael

    AU - Casacuberta, Francisco

    AU - Martinez, Mercedes Garcia

    AU - Germann, Ulrich

    AU - Gonzalez-Rubio, Jesus

    AU - Hill, Robin L.

    AU - Koehn, Philipp

    AU - A. Leiva, Luis

    AU - Mesa-Lao, Bartolomé

    AU - Ortiz-Martnez, Daniel

    AU - Saint-Amand, Herve

    AU - Tsoukala, Chara

    PY - 2014

    Y1 - 2014

    N2 - We conducted a field trial in computer-assisted professional translation to compare Interactive Translation Prediction (ITP) against conventional post-editing (PE) of machine translation (MT) output. In contrast to the conventionalPE set-up, where an MT system first produces a static translation hypothesis that is then edited by a professional translator (hence \post-editing"), ITP constantly updates the translation hypothesis in real time in response to user edits. Our study involved nine professional translators and four reviewers working with the web-based CasMaCat workbench. Various new interactive features aiming to assist the post-editor were also tested in this trial. Our results show that even with little training, ITP can be as productive as conventional PE in terms of the total time required to produce the final translation. Moreover, in the ITP setting translators require fewer key strokes to arrive at the final version of their translation.

    AB - We conducted a field trial in computer-assisted professional translation to compare Interactive Translation Prediction (ITP) against conventional post-editing (PE) of machine translation (MT) output. In contrast to the conventionalPE set-up, where an MT system first produces a static translation hypothesis that is then edited by a professional translator (hence \post-editing"), ITP constantly updates the translation hypothesis in real time in response to user edits. Our study involved nine professional translators and four reviewers working with the web-based CasMaCat workbench. Various new interactive features aiming to assist the post-editor were also tested in this trial. Our results show that even with little training, ITP can be as productive as conventional PE in terms of the total time required to produce the final translation. Moreover, in the ITP setting translators require fewer key strokes to arrive at the final version of their translation.

    KW - CAT

    KW - SMT

    KW - Interactive translation prediction

    KW - Post-editing

    KW - Field trial

    KW - User studies

    KW - CAT

    KW - SMT

    KW - Interactive translation prediction

    KW - Post-editing

    KW - Field trial

    KW - User studies

    U2 - 10.1007/s10590-014-9157-9

    DO - 10.1007/s10590-014-9157-9

    M3 - Journal article

    VL - 28

    SP - 217

    EP - 235

    JO - Machine Translation

    JF - Machine Translation

    SN - 0922-6567

    IS - 3-4

    ER -