DM_NLP at SemEval-2018 Task 8: neural sequence labeling with linguistic features

Abstract

This paper describes our submissions for SemEval-2018 Task 8: Semantic Extraction from CybersecUrity REports using NLP. The DM NLP participated in two subtasks: SubTask 1 classifies if a sentence is useful for inferring malware actions and capabilities, and SubTask 2 predicts token labels ("Action", "Entity", "Modifier" and "Others") for a given malware-related sentence. Since we leverage results of Subtask 2 directly to infer the result of Subtask 1, the paper focus on the system solving Subtask 2. By taking Subtask 2 as a sequence labeling task, our system relies on a recurrent neural network named BiLSTM-CNN-CRF with rich linguistic features, such as POS tags, dependency parsing labels, chunking labels, NER labels, Brown clustering. Our system achieved the highest F1 score in both token level and phrase level.

Deep Transformer-based Representation for Text Chunking

“…To benefit from the information of both sides of a sentence, BiLSTM was introduced by Graves et al [11], enabling models to capture more information during training which achieved better results in the chunking task. Ma et al [23] and Huang et al [15] used BiLSTM to obtain word representations with respect to both right and left context and a subsequent CRF layer to consider sentence level tag information. Huang et al [15] also used SENNA [6] pre-trained embeddings.…”

“…Huang et al [15] also used SENNA [6] pre-trained embeddings. Moreover, in the proposed model of Ma et al [23], a max-pooling and a convolutional layer were used to obtain character embeddings for each word. They also used the concatenation of character representations, linguistic features like POS and NER labels, and word embeddings to create a general embedding before feeding to BiLSTM.…”

Devising Malware Characteristics using Transformers

“…DM-NLP [10] used the predicted output labels from SubTask 2 to get the predictions for SubTask 1. They model this task as a sequence labelling task and used a hybrid approach with BiLSTM-CNNCRF as mentioned in [11].…”

DM_NLP at SemEval-2018 Task 12: A Pipeline System for Toponym Resolution

“…Finally, a voting method is utilized to benefit from multiple models. Input Information Based upon our previous work (Ma et al, 2018) on sequence labeling, our system incorporates four types of linguistic information: Part-of-Speech (POS) tags, NER labels, Chunking labels and ELMo (Peters et al, 2018). The former three are generated by open source tools.…”