Excel is a unique kind of document that contains both structured and unstructured information in a single file. Parsing it can be difficult because we need to identify the schema and locate where the structured data is located. Additionally, Excel spreadsheets can have comments, which may need to be ignored depending on the requirements.

Answer Server, a component of the OpenText IDOL suite of AI products, has recently introduced the ability to integrate with third-party LLM (Large Language Models) in its 24.1 version. To perform reasoning in the given context, a suitable model must be chosen. For this purpose, Mistral 7B was chosen, as it ranks second among all models available through an API and provides top-tier reasoning capabilities.

Extracting the context and sending the relevant information has been a challenging task. However, IDOL's KeyView can be used for content and metadata extraction. Once the content has been extracted, the Answer Server can identify the context that matches the user's question. The question and context can then be passed to the LLM to provide the answer.

In the image below, we have an excel spreadsheet containing the names of companies and response times along with SLAs. We are asking, "Who has a shorter response time in the SLA spreadsheet, CloudFrontier or InnovateIT?". IDOL can identify the intent of the question, extract both response times, and then provide the company with the shorter response time.

For a deeper understanding of these advancements or to talk about potential collaborations, please feel free to explore Opentext's overview on AI text analytics or drop me an email at vjoseph@opentext.com.  If you are based across Europe or North America please drop a line to lobrien@opentext.com

Join OpenText on LinkedIn and follow @OpenText on X.

We’d love to hear your thoughts on this blog. Comment below.

The OpenText Analytics & AI team