Researcher

Evangelos Kanoulas

Evangelos Kanoulas studies how information-retrieval systems find and use evidence. His work connects reliable evaluation with retrieval methods for language models and agents. Kanoulas completed his Northeastern computer-science doctorate in the period 2004 to 2009. His dissertation addressed reliable test and training collections in information retrieval. His work proposes a system for finding suitable expert language agents and combining their responses. The system includes an evaluation framework using thousands of synthetic expert agents, and he proposes a future ecosystem. His work includes research on negation and set operations in neural retrieval, query decomposition for retrieval-augmented generation and reinforcement learning for conversational retrieval decisions.

Works
259
Citations
4,242
h-index
35

Publication numbers from OpenAlex, as of 14 October 2026.

Photo of Evangelos Kanoulas

Timeline

  1. 2004–2009

    Reliable retrieval evaluation

    Kanoulas completed his Northeastern computer-science doctorate in the period 2004 to 2009. His dissertation addressed reliable test and training collections in information retrieval.

  2. As of October 2026

    Agent-centred information access

    He proposes a system for finding suitable expert language agents and combining their responses. He proposes an evaluation framework using thousands of synthetic expert agents and a future ecosystem.

  3. As of October 2026

    Retrieval that handles logical queries

    His work includes research on negation and set operations in neural retrieval. He also studies query decomposition for retrieval-augmented generation and reinforcement learning for conversational retrieval decisions.

Sources and credits

Photo: ellis.eu; Evangelos Kanoulas profile photograph; creator not stated. License: Not stated by source. Source: ellis.eu. Cropped to a square and resized. Ask us to remove or replace this photo.

  1. Evangelos Kanoulas - Personal Website staff.fnwi.uva.nl. Checked 13 October 2026.