Describes a prototype that extracts structured information and timelines from probation documents using PDF extraction, text mining, and language analytics. Focuses on mental health, substance use, and treatment histories to make narrative records more accessible for analysis.

Introduces computational linguistics and NLP, examining their relationship to formal linguistic theory. Reviews Persian language technologies and challenges involving the writing system, word boundaries, compounds, and light verb constructions, and identifies directions for further development.

Proposes that Persian light verbs are underspecified regular verbs and that complex predicates reproduce verb phrase structure at a smaller scale. Differences in interpretation emerge from the verb's degree of underspecification and its combination with the nonverbal element.

Analyzes the Eastern Armenian auxiliary's placement in neutral and focused sentences. Proposes second-position clisis within the verb phrase and a phase-based account of its distribution, drawing on parallels between verbal and clausal structure.

Compares collaboration patterns in theoretical linguistics and NLP using publication data, coauthorship networks, and topic analysis. The study finds a more interconnected NLP network, while the linguistics network contains many isolated authors and smaller collaboration groups, with differences across research areas and institutions.

Develops and evaluates automatic event classification for Japanese and French aviation incident reports. Uses annotated corpora to compare a Bayesian logistic regression approach with a new feature selection method against an existing classification approach.

Presents an agent-based model for studying community resilience during an epidemic. Integrates social networks, movement, and perceived contact risk to explore relationships among community responses, fatalities, and the disruption of social ties.

Examines emotional ratings of words across 24 corpora in ten languages, finding a consistent positivity bias in the sampled vocabulary. The study also develops methods for measuring the emotional content of large text collections.

Describes an evaluation of eleven language identification tools on short texts across thirty language and script combinations. Assesses language coverage, accuracy, precision and recall, speed, scalability, and robustness.

Tests Zipf’s Law across Persian corpora at different levels of linguistic analysis, including word forms, compounds, morphology, and word senses. Finds Zipfian patterns across several word-level representations, but not in the frequency distribution of syntactic categories alone.

Maps research trends and collaboration networks in ACL and NAACL publications from 2007-2014. Topic and co-authorship analyses examine the prominence of statistical machine translation, emerging semantic approaches, and the concentration and interconnectedness of research activity.

Examines how spelling, capitalization, diacritics, and character encoding affect text matching across languages. Illustrates the benefits and limitations of normalization through its application to Arabic sentiment analysis in the Hedonometer project.