Joint prediction in MST style discourse parsing for argumentation mining Andreas Peldszus Manfred Stede Applied Computational Linguistics, University of Potsdam EMNLP 20.09.2015 Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 1 / 37
Outline 1 Argumentation Mining 2 Dataset & Scheme 3 Models 4 Results for the attachment task 5 Results for all tasks Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 2 / 37
Outline 1 Argumentation Mining 2 Dataset & Scheme 3 Models 4 Results for the attachment task 5 Results for all tasks Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 3 / 37
What is argumentation mining? Health insurance companies should naturally cover alternative medical treatments. Not all practices and approaches that are lumped together under this term may have been proven in clinical trials, yet it's precisely their positive effect when accompanying conventional 'western' medical therapies that's been demonstrated as beneficial. Besides many general practitioners offer such counselling and treatments in parallel anyway - and who would want to question their broad expertise? Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 4 / 37
What is argumentation mining? [e1] Health insurance companies should naturally cover alternative medical treatments. [e2] Not all practices and approaches that are lumped together under this term may 2 have been proven in clinical trials, 1 c2 [e3] yet it's precisely their 3 c3 positive effect when accompanying conventional 'western' medical therapies that's been demonstrated as beneficial. [e4] Besides many general practitioners offer such 4 c4 counselling and treatments in parallel anyway - [e5] and who would want to 5 question their broad expertise? Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 4 / 37
What is argumentation mining? [e1] Health insurance companies should naturally cover alternative medical treatments. Tasks: [e2] Not all practices and approaches that are lumped together under this term may • EDU segmentation 2 have been proven in clinical trials, • ADU segmentation 1 c2 [e3] yet it's precisely their resp. argumentative relevance 3 c3 positive effect when accompanying conventional • ADU type classification 'western' medical therapies that's been demonstrated as beneficial. • Relation identification [e4] Besides many general • Relation type classification practitioners offer such 4 c4 counselling and treatments in parallel anyway - [e5] and who would want to 5 question their broad expertise? Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 4 / 37
Outline 1 Argumentation Mining 2 Dataset & Scheme 3 Models 4 Results for the attachment task 5 Results for all tasks Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 5 / 37
Dataset: argumentative microtexts Texts: • 112 texts: collected in a controlled text generation experiment • with professional parallel translation to English • annotated with argumentation structures • see [Peldszus and Stede, to appear] • freely available, CC-by-nc-sa license • https://github.com/peldszus/arg-microtexts Properties: + about 5 segments long + each segment is arg. relevant + explicit main claim + at least one possible objection considered Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 6 / 37
Scheme Freeman’s theory, revised & slightly generalized: [e1] Health insurance companies should naturally cover alternative medical [Freeman, 1991, 2011] [Peldszus and Stede, 2013] treatments. [e2] Not all practices and • node types = argumentative role approaches that are lumped together under this term may 2 have been proven in clinical proponent (presents and defends claims) trials, 1 c2 opponent (critically questions) [e3] yet it's precisely their 3 c3 positive effect when accompanying conventional 'western' medical therapies • link types = argumentative function that's been demonstrated as beneficial. support own claims (normally, by example) [e4] Besides many general practitioners offer such 4 c4 attack other’s claims (rebut, undercut) counselling and treatments in parallel anyway - [e5] and who would want to 5 question their broad expertise? Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 7 / 37
[e1] Health insurance companies should naturally cover alternative medical treatments. [e2] Not all practices and approaches that are lumped together under this term may have been proven in clinical trials, 2 1 [e3] yet it's precisely their positive effect when 3 accompanying conventional 'western' medical therapies that's been demonstrated as beneficial. [e4] Besides many general practitioners offer such 4 counselling and treatments in parallel anyway - 5 [e5] and who would want to question their broad expertise? Preprocessing: Graph reduction [e1] Health insurance companies should naturally cover alternative medical treatments. [e2] Not all practices and approaches that are lumped together under this term may 2 have been proven in clinical trials, 1 c2 [e3] yet it's precisely their c3 3 positive effect when accompanying conventional 'western' medical therapies that's been demonstrated as beneficial. [e4] Besides many general practitioners offer such 4 c4 counselling and treatments in parallel anyway - [e5] and who would want to 5 question their broad expertise? From complex structures. . . Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 8 / 37
Preprocessing: Graph reduction [e1] Health insurance [e1] Health insurance companies should naturally companies should naturally cover alternative medical cover alternative medical treatments. treatments. [e2] Not all practices and [e2] Not all practices and approaches that are lumped approaches that are lumped together under this term may together under this term may 2 have been proven in clinical have been proven in clinical trials, trials, 1 c2 2 1 [e3] yet it's precisely their [e3] yet it's precisely their c3 3 positive effect when positive effect when 3 accompanying conventional accompanying conventional 'western' medical therapies 'western' medical therapies that's been demonstrated as that's been demonstrated as beneficial. beneficial. [e4] Besides many general [e4] Besides many general practitioners offer such 4 practitioners offer such c4 4 counselling and treatments in counselling and treatments in parallel anyway - parallel anyway - 5 [e5] and who would want to 5 [e5] and who would want to question their broad question their broad expertise? expertise? From complex structures. . . . . . to simple support-attack graphs. Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 8 / 37
Tasks tackeled in this paper: [e1] Health insurance companies should naturally cover alternative medical treatments. • attachment (at) [e2] Not all practices and approaches that are lumped 464 pairs yes, 2000 pairs no together under this term may have been proven in clinical trials, • central claim (cc) 2 1 [e3] yet it's precisely their 112 yes, 451 no positive effect when 3 accompanying conventional 'western' medical therapies • role (ro) that's been demonstrated as beneficial. 451 proponent, 125 opponent [e4] Besides many general practitioners offer such • function (fu) 4 counselling and treatments in parallel anyway - 290 support, 174 attacks 5 [e5] and who would want to question their broad expertise? Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 9 / 37
Tasks tackeled in this paper: [e1] Health insurance companies should naturally cover alternative medical treatments. • attachment (at) [e2] Not all practices and approaches that are lumped 464 pairs yes, 2000 pairs no together under this term may have been proven in clinical trials, • central claim (cc) 2 1 [e3] yet it's precisely their 112 yes, 451 no positive effect when 3 accompanying conventional 'western' medical therapies • role (ro) that's been demonstrated as beneficial. 451 proponent, 125 opponent [e4] Besides many general practitioners offer such • function (fu) 4 counselling and treatments in parallel anyway - 290 support, 174 attacks 5 [e5] and who would want to question their broad expertise? Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 9 / 37
Tasks tackeled in this paper: [e1] Health insurance companies should naturally cover alternative medical treatments. • attachment (at) [e2] Not all practices and approaches that are lumped 464 pairs yes, 2000 pairs no together under this term may have been proven in clinical trials, • central claim (cc) 2 1 [e3] yet it's precisely their 112 yes, 451 no positive effect when 3 accompanying conventional 'western' medical therapies • role (ro) that's been demonstrated as beneficial. 451 proponent, 125 opponent [e4] Besides many general practitioners offer such • function (fu) 4 counselling and treatments in parallel anyway - 290 support, 174 attacks 5 [e5] and who would want to question their broad expertise? Peldszus, Stede (Uni Potsdam) Joint prediction for argumentation mining EMNLP 2015 9 / 37
Recommend
More recommend
