Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.23659946Overview This repository provides the evaluation dataset and reproducibility materials supporting the study entitled “An AI-Driven Conversational Workflow for Real-Time Point-of-Sale Data Access.” The study investigates a WhatsApp-based conversational workflow implemented using Impresiv AI, GPT-4.1 Mini, controlled database tools, and direct read-only access to an operational MySQL Point-of-Sale (POS) database. Dataset Contents The repository includes 50 natural-language text-test scenarios and 10 Excel-export scenarios, recorded chatbot responses, ground-truth aggregate values, evaluation scores, and workflow execution times. It contains CSV and Excel datasets, an evaluation protocol, a data dictionary, analysis summaries, and Python scripts for reproducing the reported evaluation metrics. Evaluation Summary The recorded experiments achieved: Text response correctness: 49/50 (98.0%) Manual intent/handling agreement: 49/50 (98.0%) Workflow completion: 60/60 (100%) Reported Excel generation and delivery: 10/10 (100%) Mean text workflow response time: 5.802 seconds Mean Excel workflow response time: 56.498 seconds Data Availability and Limitations The repository contains sanitized evaluation records and aggregate reference values. The original operational POS database, customer-level transaction records, database credentials, private report exports, and complete SQL execution traces are not publicly distributed due to privacy and commercial confidentiality considerations. The provided scripts reproduce the reported evaluation summaries from recorded observations. They do not rerun the live AI workflow or independently verify the contents of the generated Excel reports. Research Context The evaluation was conducted using an operational POS environment containing 12,848 transactions and 37,543 sold items during the observation period from January 1 to July 1, 2026. This repository supports transparency and reproducibility of the recorded evaluation and is not intended to serve as a general-purpose POS transaction dataset.
Keyword(s):
Conversational Interface; Evaluation Dataset; Impresiv AI; Point-of-Sale; Read-Only SQL.
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
No Associated Publication Data Available
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)