Skip to content

This repository contains the experiments to the AAMAS 2019 paper.

License

Notifications You must be signed in to change notification settings

mir-pucrs/norm-conflict-classification

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

43 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Norm Conflict Classification

This repository contains the code used in the experiments of the paper entitled "Classification of Contractual Conflicts via Learning of Semantic Representations" accepted in the International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2019) by João Paulo Aires, Roger Granada, Juarez Monteiro, Rodrigo C Barros and Felipe Meneguzzi.


Abstract

Contracts are the main medium through which parties formalize their trade relations, be they the exchange of goods or the specification of mutual obligations. While electronic contracts allow automated processes to verify their correctness, most agreements in the real world are still written in natural language, which need substantial human revision effort to eliminate possible conflicting statements in long and complex contracts. In this paper, we formalize a typology of conflict types between clauses suitable for machine learning and develop techniques to review contracts by learning to identify and classify such conflicts, facilitating the task of contract revision. We evaluate the effectiveness of our techniques using a manually annotated contract conflict corpus with results close to the current state-of-the-art for conflict identification, while introducing a more complex classification task of such conflicts for which our method surpasses the state-of-the art method.


Conflict Identification and Classification

We develop two approaches for norm conflict identification, i.e., classify unseen pairs of norms as conflict or non-conflict, and two further approaches to classify the type of conflict that occurs between norm pairs (deontic-modality, deontic-structure, deontic-object, or object-conditional). Before identifying or classifying norms, we transform each norm written in natural language within a contract into a vector representation using sent2Vec.

Our approach to identify norm conflicts are based on generating comparative embeddings by computing the distance between norm embeddings Embnorm and using these distances as a semantic representation of the presence or absence of norm conflicts (i.e., conflicts and non-conflicts). We developed two such distance-based approaches to find the comparative embedding corresponding to the conflicts between norms, both of which are based on finding the mean value (centroid in embedding space). The first approach consists of identifying the centroid of concatenated embeddings of norm pairs representing conflicts and the centroid of concatenated embeddings representing non-conflicts. The centroid is generated by computing the mean of this extended embedding space, which we call Embcent. The second approach consists of computing two centroids for the offset embeddings of norm pairs. The centroid is generated by computing in the same embedding space as the original norms

$Emb_{cent} = \frac{\sum_{i=1}^{C} Emb_{off}{[i]}{C}$

where

$$Emb_{off} = Emb_{norm}[1] - Emb_{norm}[2]$$

represents the offset generated from two norms n1 and n2 of the set C.

Alt text


How to cite

When citing our work in academic papers, please use this BibTeX entry:

@inproceedings{AiresEtAl2018aamas,
  author    = {Aires, Jo\~{a}o Paulo and Granada, Roger and Monteiro, Juarez and Barros, Rodrigo C and Meneguzzi, Felipe},
  title     = {Classification of Contractual Conflicts via Learning of Semantic Representations},
  booktitle = {Proceedings of the 18th International Conference on Autonomous Agents and Multiagent Systems},
  series    = {AAMAS 2019},
  location  = {Montreal, Canada},
  month     = {May},
  year      = {2019},
  publisher = {International Foundation for Autonomous Agents and Multiagent Systems}
}