Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Database design and smart matching of universities and businesses for technology transfer offices

Papadogiannis Dimitrios

Full record

---

Summary

Every higher-education institution must support its academic staff and highlight all the innovative products and services created or provided by its members. The Technology Transfer Office of the Technical University of Crete undertakes precisely this interconnection of the Institution with businesses and other bodies. However, in today's era of digital transformation, for the efficient operation of the Office there should also be support from suitable digital tools that collect, analyze and retrieve data effectively, facilitating information exchange and communication between users. Towards this end, technologies from the domain of Web Application Programming are utilized in combination with the closely-linked domain of Databases. The purpose of this diploma thesis is the design and implementation of an integrated online support application for the Technology Transfer Office of the Technical University of Crete. The application registers information about the Schools' Laboratories, the Innovative Products of the Inventors and the data of interested Businesses. The application then provides methods for search and intelligent data retrieval, in order to facilitate the smart matching of university and businesses in cases of common interest. Open source tools were used to implement the application. A well-structured MariaDB database and a dynamic website were created, using the PHP Framework Laravel, which follows the Model-View-Controller architecture and offers a responsive environment for a wide range of fixed and mobile devices. One of the main innovative elements of the application is the advanced search engine that combines two approaches: (a) the classic search implemented through SQL queries to find records where specific keywords appear, without taking into account any surrounding information (context-free) and (b) a smarter mechanism based on natural language processing and artificial intelligence, using the pre-trained GreekBERT model in combination with KeyBERT to extract context-aware keywords. This mechanism, which utilizes methods, such as text segmentation and the creation of n-grams, generates a score for each possible keyword and applies filtering to reject keywords that appear in a negative context. Thus, the accuracy of information retrieval is significantly improved. Finally, the application provides, through an Administrator role (backend), the ability to manage users and records with full access to creation, modification, deletion and configuration. User evaluation of the application demonstrates that the application functions satisfactorily, offering users a fast and accurate search engine that understands the context of the data, also laying the foundation for further improvements and extensions in the future.

Available Files

Services