What represents “style” in authorship attribution?

Proceedings of the 27th International Conference on Computational Linguistics, pages 2814–2822 Santa Fe, New Mexico, USA, August 20-26, 2018. 2814 What constitutes “style” in authorship attribution? Kalaivani Sundararajan Florida Institute for Cybersecurity University of Florida kalaivani.s@ufl.edu Damon L. Woodard Florida Institute for Cybersecurity University of Florida dwoodard@ece.ufl.edu Abstract Authorship attribution typically uses all information representing both content and style whereas attribution based only on stylistic aspects may be robust in cross-domain settings. This paper analyzes different linguistic aspects that may help represent style. Specifically, we study the role of syntax and lexical words (nouns, verbs, adjectives and adverbs) in representing style. We use a purely syntactic language model to study the significance of sentence structures in both single- domain and cross-domain attribution, i.e. cross-topic and cross-genre attribution. We show that syntax may be helpful for cross-genre attribution while cross-topic attribution and single-domain may benefit from additional lexical information. Further, pure syntactic models may not be effective by themselves and need to be used in combination with other robust models. To study the role of word choice, we perform attribution by masking all words or specific topic words corresponding to nouns, verbs, adjectives and adverbs. Using a single-domain dataset, IMDB1M reviews, we demonstrate the heavy influence of common nouns and proper nouns in attribution, thereby highlighting topic interference. Using cross-domain Guardian10 dataset, we show that some common nouns, verbs, adjectives and adverbs may help with stylometric attribution as demonstrated by masking topic words corresponding to these parts-of-speech. As expected, it was observed that proper nouns are heavily influenced by content and cross-domain attribution will benefit from completely masking them. 1 Introduction Authorship attribution attributes a text sample to a person based on either content or their writing style. On the other hand, stylometry involves authorship attribution only based on writing style and needs to be independent of topic or genre. Though authorship attribution has various applications, some important ones are security-related. eg. cybercrime investigation, digital forensics, countering identity deception in social networks, etc. In today’s connected world, most of our cyber-interactions are text-based on various platforms like messages, emails, blogs, tweets, forum posts etc. With rampant cybercrime attacks, it is important to perform robust authorship attribution or verification across these platforms. This calls for stylometric attribution approaches that model style and work well in cross-domain scenarios. Cross- domain scenarios include both cross-topic (same genre but different topics) and cross-genre (different genres) scenarios. Various authorship attribution approaches have been surveyed in literature (Juola and others, 2008; Stamatatos, 2009; Neal et al., 2017). However, only some of these approaches focus specifically on cross- domain attribution. Function words have been touted as content-independent and hence reliable as a style marker (Juola and others, 2008). Syntactic information, both shallow (Luyckx and Daelemans, 2008) and deep (Baayen et al., 1996; Raghavan et al., 2010; Feng et al., 2012; Fuller et al., 2014; Bj¨orklund and Zechner, 2017), have also been proposed for ensuring topic or genre independence. Few other approaches, perform preprocessing to prevent topic/genre dependency when …

Content truncated to 500 words