Client and Challenge:
Rebold, part of the ISPD group, is a leader in communication solutions and data analysis. The Rebold-ISPD Oliva Project focused on the digitalization and automation of content enrichment through artificial intelligence, aiming to reduce human intervention. The key challenges included the annotation and reporting of large volumes of text and audiovisual content, with a focus on improving accuracy and efficiency.
Project Overview:
Dive collaborated with Rebold-ISPD to review and improve the annotation and reporting processes of the Oliva Project. The goal was to develop AI models to automate these processes, increasing efficiency and accuracy across various use cases, while reducing human intervention. The analyzed use cases included Named Entity Recognition (NER), sentiment evaluation, prominence, and news grouping by themes, among others.
Implemented solutions:
Dive conducted a thorough consultancy and analysis of Rebold-ISPD’s current processes, focusing on reviewing the developments made by Gradiant, the entity contracted for the AI model development within the Oliva project. Use cases were classified by priority and complexity, and specific requirements for each were defined.
In the design phase, appropriate model architectures were selected, and relevant datasets for training and validation were collected and curated. The developed AI models included Deep Learning techniques and algebraic operations in the latent space of embeddings, enabling precise and efficient classification of informational units.
Technology for specialized services:
The models were deployed on ISPD’s infrastructure, using large language models (LLMs) to generate embeddings and classify the informational units. Neural networks were implemented for classification tasks and LLMs for specific applications requiring advanced cognitive capabilities.
Achievements in AI Annotation for Rebold
Achievements:
The implementation of Dive’s AI models resulted in a significant improvement in the accuracy and efficiency of Rebold-ISPD’s annotation and reporting processes. The obtained data and statistics demonstrated a reduction in processing time, increased accuracy in entity detection and classification, and improved customer satisfaction by providing more detailed and accurate reports.
Why Dive?
Dive stands out for its innovative and customized approach in applying artificial intelligence to solve complex challenges. Our mission is to transform our clients’ operational processes, maximizing efficiency and productivity through advanced AI solutions tailored to the specific needs of each project.
Thanks to the collaboration with Dive, we have significantly automated and improved our annotation and reporting processes. The achieved precision and efficiency allow us to offer better service to our clients and optimize our internal resources.
