What we say reflects who we are. Natural language processing allows us to explore and exploit this strong relationship between language and the humans who produce it, with applications both within the field and across many social science disciplines. Understanding who we are, such as our personality or well-being, requires models that are both accurate and interpretable, particularly for an interdisciplinary audience. Understanding language based on who we are requires both a suitable author representation and a model that is capable of leveraging it.
In this work, we seek to answer the following research questions: (1) how important is the relationship between humans and language for improving NLP tasks and methods, (2) how can we use language to better model human attributes, and (3) how can we create NLP models that can leverage knowledge of the author? We present a neural model for personality prediction that uses message-level attention to identify meaningful messages, leading to more interpretable results. We also introduce user-factor adaptation, a technique for creating human-aware models that outperform human-agnostic ones. This work is a step towards more “human-centered” NLP, advocating for modeling that is more accurate, more interpretable, and more personal.
|Advisor:||Balasubramanian, Niranjan, Schwartz, Hansen A|
|Commitee:||Skiena, Steven, Samaras, Dimitris, Hovy, Dirk|
|School:||State University of New York at Stony Brook|
|School Location:||United States -- New York|
|Source:||DAI-B 81/9(E), Dissertation Abstracts International|
|Keywords:||Language modeling, Natural language processing, User factors|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be