CAREER: Authorship Analysis in Cross-Domain Settings

作者分析（AA）是从书面文档中提取特征的任务，这些特征可以帮助确定文档的作者身份，生成作者的个人资料，或识别剽窃案例。AA可以用于历史目的，解决对给定文件的原始创建者的争议，以及对网络滥用者提起诉讼。在AA中，大多数先前的工作假设具有已知作者的样本的可用性，这些样本与感兴趣的文档的领域密切匹配。像这样一个强有力的假设限制了AA方法的应用。该项目通过为不同的跨领域程度的场景设计健壮的框架来解决这一关键的突出挑战：跨主题、跨体裁和跨情态（文本与转录语音）。该项目利用大量可用的自由文本来表示每个跨域设置，以学习一般的词汇和句法分布对应。这些对应关系用于将域外文本映射到更接近目标域的表示。本研究的直接贡献包括以分布轨迹的形式提取和嵌入跨域先验知识到AA模型中的新方法；以及对主题、体裁和情态在AA特征工程过程中的影响的深刻理解，这也将有助于其他文本处理任务。这项研究将对与国家安全密切相关的法律语言学领域做出直接贡献。PI将设计一个关于法律语言学计算方法的高级研讨会，并将为STEM学科中代表性不足的群体扩大她正在进行的教育和外展活动。该项目将为参与该项目的研究生提供国际重点研究实验室访问的机会，这将丰富他们的培训并提供良好的交流机会。

Authorship Analysis (AA) is the task of extracting characteristics from written documents that can help to determine authorship of a document, generate a profile of the author, or identify cases of plagiarism. AA can be used for historical purposes, to settle disputes over the original creators of a given document, and to build a prosecution case against an online abuser. Most previous work in AA assumes the availability of samples with known authorship that closely match the domain of the documents of interest. A strong assumption like this one limits the applications of AA approaches. This program addresses this key outstanding challenge by designing robust frameworks for scenarios with different cross-domain degrees: cross-topic, cross-genre and cross-modality (text vs. transcribed speech). The project leverages the large amounts of free text available representing each cross-domain setting to learn general lexical and syntactic distributional correspondences. These correspondences are used to map the out-of-domain texts to a representation that is closer to the target domain. Direct contributions of this research include new approaches to extract and embed cross-domain prior knowledge into AA models in the form of distributional trajectories; and a solid understanding of the influence of topic, genre, and modality in the feature engineering process for AA that will also be helpful in other text processing tasks. This research will make direct contributions to the field of forensic linguistics, which is of major relevance for national security.The PI will design an advanced seminar in computational approaches for forensic linguistics and will expand her ongoing educational and outreach activities for underrepresented groups in the STEM disciplines. The PI will integrate opportunities for international visits to key research labs for the graduate students involved in the program that will enrich their training and provide great networking opportunities.