Isabella Gagliardi

She obtained a degree in Physics, with a specialization in Applied Physics and a focus on Cybernetics, in September 1985, with a score of 105/110. Her thesis was titled Managing Scientific Data: An Innovative Prototype, which was conducted at the Computer Service of the Milan Area (SIAM) at the Institute of Cosmic Physics and Related Technologies (IFCTR) of the National Research Council (CNR).

Research Activity

• Research activities in the field of IMATI mathematical informatics have been focused on designing and implementing innovative computer tools for the following purposes (since 2017):

1. Management, Analysis, Visualization, and Accessibility of Multimedia Data: This involves utilizing Data Science techniques to handle, analyze, visualize, and provide access to multimedia data, including texts, images, videos, and audio.

2. Customized Consultation and Access Paths: Research efforts have been dedicated to defining personalized consultation and access paths within digital libraries and both tangible and intangible cultural assets.

3. Methods, Models, and Tools for Automatic Text Analysis and Text/Data Mining: In the context of Natural Language Processing and employing Machine Learning methods, research has focused on automatically extracting keywords, determining semantic similarities between texts (including multilingual texts), enhancing data quality, normalizing and harmonizing terms, and identifying contextually significant terms for various applications.

 

• Previous activities have encompassed various areas, including:

1. Multimedia Systems: In this domain, topics such as data structures in automatic image indexing systems and the evaluation of the effectiveness of content search algorithms have been explored.

2. Interactive Multimedia Information Systems on the Internet: Innovative tools for visualization, accessibility, and searching of multimedia data have been developed and applied to accessible internet-based multimedia information systems.

3. Multimedia Information Systems Applied to Intangible Cultural Assets: These systems have been extended to intangible cultural assets, broadening their applicability.

4. Customizable Collaborative Platforms for Managing Project/Legislative Documents and Related Multimedia Attachments: These platforms encompass both back-end and front-end components, as well as user management tools, search capabilities, browsing functionalities, and facilitate the handling of multimedia attachments.

5. Collaborative Systems for Managing Terminological Resources: Integration of terminological resources with the developed information systems and with publicly available databases on the internet has been a focus area.

6. Linked Open Data Technologies Applied to Information Systems: The integration of Linked Open Data technologies into the developed information systems has been explored to enhance data interoperability and accessibility.

7. Tourism: Research efforts have been dedicated to developing methods, techniques, and practical tools for promoting tourism as a sustainable, high-quality, and profitable productive system, integrated with related industrial sectors.

 

 

 

Textual Semantic Similarity and Multilevel Semantic Graphs

The work of managing cultural assets has naturally led us to develop methods and tools for the automatic identification of fundamental concepts within a text and for a visualization that is both simple, captivating, and informative. Our research activities in the realm of text mining are dedicated to the exploration and refinement of sophisticated methodologies and algorithms, all geared towards the extraction of valuable insights and knowledge from textual data sources. Our research also delves into the domain of data visualization, where we strive to create innovative approaches and tools for presenting complex information in a visually comprehensible manner, aiding in the effective communication and interpretation of data-driven insights.

• Semantic Similarity of Textual Data from Various Sources: I aim to continue my research in the field of Natural Language Processing related to semantic similarity between short texts and keywords. The initial results will be integrated into QueryLab. New experiments will be conducted on textual data with the goal of defining new language-independent automatic query expansion tools. To achieve this, various techniques and methods will be integrated, including machine learning tools, transformers, LTSM, word embeddings, and those based on hierarchical and semantic word organizations, such as WordNet, to overcome language barriers.

• Automatic Creation of Multilevel Semantic Graphs: Initial results for the unsupervised creation of semantic graphs have been achieved by integrating transformers, dimensionality reduction algorithms like UMAP, and clustering algorithms like HDBSCAN. The experimentation has been based on QueryLab's inventory of intangible assets, providing multilevel semantic graphs for content navigation and visualization. I intend to continue these activities by addressing the main challenges: proper identification of hyperparameters for the key algorithms used and selecting the most suitable pre-trained language model for the purpose and dataset. I also plan to develop tools to assist users in navigating the graphs to improve usability and the user experience, especially for less experienced users.

  The paper Intuitive Semantic Graph Tool for Enhanced Archive Exploration has been published in ACM Hypertext 2023 conference.

  -

Activities carried out in the Lab

Multimedia Information Systems activities at IMATI-MI

Projects

Cleverness

More info...

DIGIDOCS

More info...

Geca, a complete cataloguing system of IMATI -CNR

More info...

Portale delle Fonti della Repubblica Italiana

More info...

PRIN 2022 ARDIPS

More info...

PRIN PNRR SMILE

More info...

Project DigitXL

More info...

QueryLab: search engine for Cultural Heritage

More info...

Geca

Portale delle Fonti

PRIN 2022 PNRR SMILE

Sistema Gestionale

Other Projects

I-ZEB

Publications