Returning the N to NLP: Towards Contextually Personalized Classification Models

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7828–7838 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 7828 Returning the N to NLP: Towards Contextually Personalized Classification Models Lucie Flek Mainz University of Applied Sciences Germany lucie.flek@hs-mainz.de Abstract Most NLP models today treat language as universal, even though socio- and psycholin- gustic research shows that the communicated message is influenced by the characteristics of the speaker as well as the target audience. This paper surveys the landscape of personal- ization in natural language processing and re- lated fields, and offers a path forward to miti- gate the decades of deviation of the NLP tools from sociolingustic findings, allowing to flex- ibly process the “natural” language of each user rather than enforcing a uniform NLP treat- ment. It outlines a possible direction to incor- porate these aspects into neural NLP models by means of socially contextual personaliza- tion, and proposes to shift the focus of our eval- uation strategies accordingly. 1 Introduction Our language is influenced by one’s individual char- acteristics as well as by the affinity to various so- ciodemographic groups (Bucholtz and Hall, 2005; McPherson et al., 2001; Eckert and McConnell- Ginet, 2013). Yet the majority of NLP models today treats language as universal, acknowledging that words have different meanings in different se- mantic context, but typically assuming that this context has the same meaning for everyone. In this paper, I propose that our focus shifts towards interpreting the language together with its user- dependent, contextual personal and social aspects, in order to truly process the “natural” language of a user. I outline a possible direction to incorporate these aspects into neural NLP models, and suggest to adjust our evaluation strategies. The paper is structured with the following aims in mind: Sec. 2 provides historical context, seeking evidence on personalization needs. Sec. 3 reviews existing personalization work, as the personaliza- tion efforts and success stories are scattered across contributions to various applied tasks. Sec. 4 con- templates on how NLP personalization could be adopted as a process of several stages. Sec. 5 out- lines an implementation proposal on contextually personalized classification models, building upon flexible, socially conditioned user representations. Sec. 6 proposes novel evaluation approaches re- flecting the benefit of personalized models. Finally, Sec. 7 opens the discussion on ethical aspects, non- personalizable NLP tasks, and the role of industry in personal data collection and protection. 2 Historical context Since 1990s, with the rise of so-called empirical or statistical NLP area (Manning et al., 1999; Brill and Mooney, 1997), the focus on frequently appearing phenomena in large textual data sets unavoidably led to NLP tools supporting “standard English” for generic needs of an anonymous user. An NLP tool - whether e.g. a POS tagger, dependency parser, ma- chine translation model or a topic classifier - was typically provided as one trained model for one lan- guage (Toutanova et al., 2003; Klein and Manning, 2003; Morton et al., 2005), or, later on, for …

Content truncated to 500 words