TACAM: Topic And Context Aware Argument Mining

Abstract

In this work we address the problem of argument search. The purpose of argument search is the distillation of pro and contra arguments for requested topics from large text corpora. In previous works, the usual approach is to use a standard search engine to extract text parts which are relevant to the given topic and subsequently use an argument recognition algorithm to select arguments from them. The main challenge in the argument recognition task, which is also known as argument mining, is that often sentences containing arguments are structurally similar to purely informative sentences without any stance about the topic. In fact, they only differ semantically. Most approaches use topic or search term information only for the first search step and therefore assume that arguments can be classified independently of a topic. We argue that topic information is crucial for argument mining, since the topic defines the semantic context of an argument. Precisely, we propose different models for the classification of arguments, which take information about a topic of an argument into account. Moreover, to enrich the context of a topic and to let models understand the context of the potential argument better, we integrate information from different external sources such as Knowledge Graphs or pre-trained NLP models. Our evaluation shows that considering topic information, especially in connection with external information, provides a significant performance boost for the argument mining task.

Relational and Fine-Grained Argument Mining

“…As the experiments in our previous work ( [9], see also Table 1) showed, there is still a huge gap of 16% Macro-F 1 score between the two-class and the three-class crosstopic scenario and of 8% in the in-topic scenario. The reason is that stance detection is a complex task.…”

“…In previous work [9], some of us addressed the problem of topic-focused argument extraction on the sentence-level. Examples of the type of sentences that we extract can be seen in Fig.…”

“…Our experimental results show that considering topic and context information from pre-trained models improves upon state-of-the-art argument detection models considerably. The number of parameters of the models and the hyper parameters of the training are reported in the previous publication [9].…”

Diversity Aware Relevance Learning for Argument Search

“…We investigate two approaches to obtain vector representations on which we compute similarities using l1, l2 or cos similarity. Previous works demonstrated that BERT models pre-trained on the task of language modeling can capture argumentative context [10]. Thus, our first BERT similarity function employs a BERT model without fine-tuning to encode the premises.…”

“…The query can be defined as a topic, e.g. Energy in which case the ARS retrieves all possible arguments without further specification [10,15,17]. Our work deals with a more advanced case, where a query is formulated in form of a claim and the user expects premises attacking or supporting this query claim.…”