基于形式论辩系统的滑坡论证分析——人工智能如何理解自然语言论辩
|
摘要
形式论辩是当前人工智能领域的研究热点之一。作为一种非单调推理形式手段,它可以与自然语言处理、论证挖掘等多个研究方向结合。然而,如何运用形式论辩理论来建模非形式论证,仍然是一个开放性问题。文章以此为出发点,着眼于一种在法律、伦理领域常见的非形式论证——滑坡论证,使用形式论辩方法对其进行建模。借鉴非形式逻辑领域论证图式的研究,文章对一个源于自然语言文本的滑坡论证案例,即"明日复明日",进行了深入分析。在此基础上结合形式论辩领域中的抽象论辩框架及结构化论辩ASPIC+系统,对滑坡论证进行刻画,构建了一个针对滑坡论证的形式化论辩理论。案例研究显示,使用文章给出的滑坡论辩理论推理的结果符合人类推理直觉。
As an approach for non-monotonic reasoning,formal argumentation is a growingly popular research area in Artificial Intelligence. It inspires a couple of other research areas in AI,such as nature language processing,argument mining,etc. One of the most important problem in the study of argumentation is how to bridge the gap between formal argumentation systems and informal argumentation studies. Focusing on the slippery slope arguments( SSA),an argument pattern that has been widely used in the fields of law and ethics,the paper tries to model natural language argument by formal argumentation. By consulting some relative researches on informal logic,for example,argumentation schemes,the paper deeply analyzes a case of SSA from natural language text,models it by structured argumentation framework ASPIC + and designs a formal argumentation theory for SSA. A case study then reveals that the results gained by using our argumentation theory for SSA are in line with human intuition.
As an approach for non-monotonic reasoning,formal argumentation is a growingly popular research area in Artificial Intelligence. It inspires a couple of other research areas in AI,such as nature language processing,argument mining,etc. One of the most important problem in the study of argumentation is how to bridge the gap between formal argumentation systems and informal argumentation studies. Focusing on the slippery slope arguments( SSA),an argument pattern that has been widely used in the fields of law and ethics,the paper tries to model natural language argument by formal argumentation. By consulting some relative researches on informal logic,for example,argumentation schemes,the paper deeply analyzes a case of SSA from natural language text,models it by structured argumentation framework ASPIC + and designs a formal argumentation theory for SSA. A case study then reveals that the results gained by using our argumentation theory for SSA are in line with human intuition.
引文
[1]TOULMIN S.The uses of argument[M].Cambridge:Cambridge University Press,2003:89-100.
[2]WALTON D,REED C,MACAGNO F.Argumentation schemes[M].Cambridge:Cambridge University Press,2008:7-42.
[3]PERELMAN CH,OLBRECHTS-TYTECA L.The new rhetoric:a treatise on argumentation[M].Notre Dame:University of Notre Dame Press,1969:19.
[4]HAMBLIN C L.Mathematical models of dialogue[J].Theoria,1971,37(2):130-155.
[5]熊明辉.非形式逻辑视野下的论证评价理论[J].自然辩证法研究,2006(12):22-25;41.
[6]谢耘.论证逻辑、非形式逻辑、论证理论[J].自然辩证法研究,2008(3):38-43.
[7]DUNG P M.On the acceptability of arguments and its fundamental role in nonmonotonic reasoning,logic programming and n-person games[J].Artificial intelligence,1995,77(2):321-357.
[8]MODGIL S,PRAKKEN H.A general account of argumentation with preferences[J].Artificial intelligence,2013,195:361-397.
[9]MODGIL S,PRAKKEN H.The ASPIC+framework for structured argumentation:a tutorial[J].Argument and computation,2014,5(1):31-62.
[10]GARCIA A,SIMARI G.Defeasible logic programming:De LP servers,contextual queries,and explanations for answers[J].Argument and computation,2014,5(1):63-88.
[11]TONI F.A tutorial on assumption-based argumentation[J].Argument and computation,2014,5(1):89-117.
[12]LIAO B,JIN L,KOONS R.Dynamics of argumentation systems:a division-based method[J].Artificial intelligence,2011(175):1790-1814.
[13]WALTON D.Slippery slope arguments[M].Oxford:Oxford University Press,1992:13-16.
[14]WALTON D.The basic slippery slope argument[J].Informal logic,2015,35(3).
[15]BARONI P,GIACOMIN M.Semantics of abstract argument systems[M]∥RAHWAN I,SIMARI G R.Argumentation in artificial intelligence.Boston:Springer,2009,chapter 2:25-44.
[16]POLLOCK J L.Defeasible reasoning[J].Cognitive science,1987,11(4):481-518.
(1)关于ASPIC+框架的详细设定,可参阅文献[8-9]。
(1)在ASPIC+中更通用的选择是采用语言L中的元素之间具有反对关系,可参见文献[9],第57页。
(2)“n(r)”表示规则r可应用。
(1)+ASPIC框架中,论辩系统的硬性规则在逆否(Transposition)或对置(Contraposition)下闭合是最终得到的结论外延满足一致性要求的必要条件,详见文献[9],第45-47页。
[2]WALTON D,REED C,MACAGNO F.Argumentation schemes[M].Cambridge:Cambridge University Press,2008:7-42.
[3]PERELMAN CH,OLBRECHTS-TYTECA L.The new rhetoric:a treatise on argumentation[M].Notre Dame:University of Notre Dame Press,1969:19.
[4]HAMBLIN C L.Mathematical models of dialogue[J].Theoria,1971,37(2):130-155.
[5]熊明辉.非形式逻辑视野下的论证评价理论[J].自然辩证法研究,2006(12):22-25;41.
[6]谢耘.论证逻辑、非形式逻辑、论证理论[J].自然辩证法研究,2008(3):38-43.
[7]DUNG P M.On the acceptability of arguments and its fundamental role in nonmonotonic reasoning,logic programming and n-person games[J].Artificial intelligence,1995,77(2):321-357.
[8]MODGIL S,PRAKKEN H.A general account of argumentation with preferences[J].Artificial intelligence,2013,195:361-397.
[9]MODGIL S,PRAKKEN H.The ASPIC+framework for structured argumentation:a tutorial[J].Argument and computation,2014,5(1):31-62.
[10]GARCIA A,SIMARI G.Defeasible logic programming:De LP servers,contextual queries,and explanations for answers[J].Argument and computation,2014,5(1):63-88.
[11]TONI F.A tutorial on assumption-based argumentation[J].Argument and computation,2014,5(1):89-117.
[12]LIAO B,JIN L,KOONS R.Dynamics of argumentation systems:a division-based method[J].Artificial intelligence,2011(175):1790-1814.
[13]WALTON D.Slippery slope arguments[M].Oxford:Oxford University Press,1992:13-16.
[14]WALTON D.The basic slippery slope argument[J].Informal logic,2015,35(3).
[15]BARONI P,GIACOMIN M.Semantics of abstract argument systems[M]∥RAHWAN I,SIMARI G R.Argumentation in artificial intelligence.Boston:Springer,2009,chapter 2:25-44.
[16]POLLOCK J L.Defeasible reasoning[J].Cognitive science,1987,11(4):481-518.
(1)关于ASPIC+框架的详细设定,可参阅文献[8-9]。
(1)在ASPIC+中更通用的选择是采用语言L中的元素之间具有反对关系,可参见文献[9],第57页。
(2)“n(r)”表示规则r可应用。
(1)+ASPIC框架中,论辩系统的硬性规则在逆否(Transposition)或对置(Contraposition)下闭合是最终得到的结论外延满足一致性要求的必要条件,详见文献[9],第45-47页。