Skip to content
Kean CS Lab

Program 05 / 06

Data, Language & Knowledge Systems

Data science is not a pile of notebooks. It is a claim about a population, a schema, and a failure mode.

The Ph.D. program names data science as a core focus. Faculty cover NLP, data mining, databases, ontologies, and time-series learning. The lab stance is conservative about hype and liberal about rigor: a language model is a tool; a knowledge graph is a hypothesis about the world; a dashboard is not a dissertation.

Questions the work is for

  1. 01

    When does an ontology earn its keep versus a fine-tuned extractor?

  2. 02

    How do we mine campus, civic, or clinical text without laundering bias into policy?

  3. 03

    What does a reproducible data-science thesis look like when the interesting data cannot leave the room?

Methods, data, benchmarks

method

NLP & information extraction

Transformer pipelines, evaluation on held-out domains, and error taxonomies (Wang, Kollapally, Li). Report the confusion, not only the F1.

method

Ontologies & knowledge graphs

Kollapally’s ontology and database work: formal vocabularies, alignment, and query. A graph that cannot answer a question a practitioner cares about is decoration.

method

Time-series & NAS

Lyu’s predictive analytics and neural architecture search — search spaces you can name, compute budgets you can defend.

tooling

Relational & web-scale data

Huang’s database and web-application stack for collecting, cleaning, and serving research data with audit trails.

Taxonomy on this page: method, dataset, benchmark, tooling. Field-standard methods are named as such; they are not automatic claims of a Kean dataset release.

Doctoral tracks

Language as infrastructure

NLP projects should pick a domain (education, health, civic records) and a decision. ‘We fine-tuned a model’ is a method paragraph, not a thesis statement.

Knowledge systems

Ontologies plus databases: the unfashionable, durable half of AI. Excellent fit for students who like formal structure and messy institutions.

Discrete foundations

Emanouilidis’s combinatorics and number theory remind the lab that algorithms have proofs. Theory-leaning doctoral work is welcome when it speaks to a computing problem.

Who this is for

Bring fluency in data wrangling and at least one of: NLP, databases, or statistical learning. You will be asked what population your numbers describe.

Ph.D. Program →

Faculty in this program

Selected papers

← All programs