i2k Connect’s Joshua Eckroth to Present on Large Language Models at SPE Data Science Convention 2024

March 15, 2024
i2k-press-release-tile

SPE Data Science Convention, to be held April 11, focuses on advanced data science solutions for any domain in the energy industry.

HOUSTON, TEXAS (March 12, 2024) — Dr. Joshua Eckroth, Chief Architect at i2k Connect, will present “How Large Language Models Help to Surface Knowledge Hidden in Vast Repositories” at the 2024 SPE Data Science Convention. His presentation will feature lessons learned about how to utilize large language models and other technologies to construct a platform that answers natural language questions about large and private corpora. These corpora are often vast, containing millions or more diverse documents, requiring sophisticated techniques to extract data, text, tables, figures, and other content.

Eckroth will discuss the deployment of “retrieval-augmented generation” (RAG), the architecture commonly used for question-answering within a corpus of documents. Complexities that are experienced in large-scale applications and that require domain-specific knowledge will be addressed. Aramco and i2k Connect have launched initiatives with SEG and SPE to build a language model like ChatGPT that is specifically trained to understand questions and construct answers about the energy field (called “Energy LLM” or “ELLM”). Eckroth will report on learnings from building domain-specific large language models such as ELLM.

An event of the SPE-GCS Data Analytics Study Group, the convention focuses on the entire life cycle of data science, from identifying the business problem, developing a solution by a multidisciplinary team, delivering the solution to the stakeholders, to sustaining the solution through monitoring. The event focuses on how to seamlessly and ethically integrate AI into oil and gas operations as an indispensable piece toward realizing efficiency, sustainability, and operational excellence.

At i2k Connect, Eckroth has primary responsibility for the software architecture that underpins the i2k Connect AI Platform. He leads research efforts such as developing Energy LLM and is an Associate Professor of Computer Science at Stetson University. He has given dozens of presentations and has authored 40 peer-reviewed publications and a book about building and deploying AI.

Download presentation

To register for DSC 2024, visit https://www.spegcs.org/events/6844/.

About i2k Connect Inc.

Founded in 2014, i2k Connect develops and deploys the i2k Connect® AI Platform. The software is game-changing AI technology that reduces by 90% the manual effort users spend to find, extract, and analyze unstructured data in documents across multiple silos of information. The i2k Connect AI Platform creates value by helping users find often buried information, leading to better decision-making, safer, more efficient operations, and quantifiable bottom-line results. At i2k Connect our goal is to partner with our customers to solve their pressing needs. How can we help? Let’s have a conversation.

Media Contact

For more information, contact Becky Thomas, +1 832-821-0866, i2kai@i2kconnect.com