BrainMap

Stefan Ruseti

- UNIVERSITATEA NAȚIONALĂ DE ȘTIINȚĂ ȘI TEHNOLOGIE POLITEHNICA BUCUREȘTI

Researcher | Teaching staff

Web of Science ResearcherID: IRZ-8715-2023

Expertise & keywords

Innovative solution to optimize user productivity through multimodal activity and profile monitoring Call name:
121491 2021 - 2023

Role in this project:

Coordinating institution: RESEARCH TECHNOLOGY SRL

Project partners: RESEARCH TECHNOLOGY SRL (); UNIVERSITATEA NAŢIONALĂ DE ŞTIINŢĂ ŞI TEHNOLOGIE POLITEHNICA BUCUREŞTI ()

Affiliation: UNIVERSITATEA NAŢIONALĂ DE ŞTIINŢĂ ŞI TEHNOLOGIE POLITEHNICA BUCUREŞTI ()

Project website: https://optimize.research-technology.ro/

Abstract:

From Archive to Canon: A Distant Reading of the Romanian Novel (1845-1947) Call name: P 4 - Proiecte de Cercetare Exploratorie, 2020
PN-III-P4-ID-PCE-2020-2690 2021 - 2023

Role in this project:

Coordinating institution: UNIVERSITATEA LUCIAN BLAGA

Project partners: UNIVERSITATEA LUCIAN BLAGA (RO)

Affiliation:

Project website: https://grants.ulbsibiu.ro/arcan/

Abstract:

To what extent did the quantitative aspect of the production of Romanian novels influence their quality? And how can one answer such question when that involves reading hundreds or even thousands of novels? The ARCAN project brings together a team of 11 specialists – the leader, 3 experienced researchers, 5 young researchers, and 2 doctoral students – whose objective is to perform a “distant reading” of a corpus of 1,567 Romanian novels published in the period spanning 1845 and 1947. The project reunites qualitative formal instruments (which primarily include genre theory) and quantitative computational tools (which rely on NLP technologies); it sets out to investigate how canonic works fare against the literary archive of the epoch. Apart from being the first Romanian (and even European) endeavor of this kind, the originality of ARCAN lies with its interdisciplinary dimension, which combines humanities and computer sciences, the size of its corpus, and the high level of granularity of the categories (microgenres) used to classify novels. Therefore, our project sets out not only to generate significant national and especially international impact, but also to produce cultural, educational and economic effects beyond the realm of the academia.

Automated Text Evaluation and Simplification Call name: P 1 - SP 1.1 - Proiecte de cercetare pentru stimularea tinerelor echipe independente
PN-III-P1-1.1-TE-2019-2209 2020 - 2022

Role in this project:

Coordinating institution: UNIVERSITATEA POLITEHNICA DIN BUCURESTI

Project partners: UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)

Affiliation: UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)

Project website: https://readerbench.com/projects/ates

Abstract:

Writing is an essential learning activity that requires both practice and experience. Writing is performed in academic environments, at workspaces, or for personal purposes, and people’s ability to clearly and concisely express their ideas in a coherent manner is an essential skill, difficult to both evaluate and improve. ATES (Automated Text Evaluation and Simplification) aims to help users improve the quality of their writing, in both English and Romanian languages, by providing immediate feedback, tailored to their writing style. Complex Natural Language Processing techniques, including deep learning models, will be used to automatically score essays relying on textual complexity indices, together with word embeddings, applied on annotated datasets of documents. In addition, textual complexity indices, combined with various features of the cohesion graph, will trigger rules to improve the text by comparison to baseline domain-specific documents. The system will also make suggestions for text simplification in order to improve its readability. For this matter, machine translation models will be trained on existing text simplification datasets, augmented with paraphrases, and ordered by readability scores. Moreover, the overall complexity of the text will be measured by automatically computing word Age of Acquisition (AoA) scores through incremental semantic models and regression analyses, in order to approximate the age when people adequately learn a word’s meaning.

Fake News in Romanian: A Joint Discursive and Computational Approach Call name: P 1 - SP 1.1 - Proiecte de cercetare pentru stimularea tinerelor echipe independente
PN-III-P1-1.1-TE-2019-1794 2020 - 2022

Role in this project:

Coordinating institution: UNIVERSITATEA LUCIAN BLAGA

Project partners: UNIVERSITATEA LUCIAN BLAGA (RO)

Affiliation: UNIVERSITATEA LUCIAN BLAGA (RO)

Project website: https://grants.ulbsibiu.ro/fakerom

Abstract:

In recent years, fake news has become both a topic of increased interest for social sciences researchers (in media studies, sociology, psychology, political sciences, etc.) and a global phenomenon with profound political, social, economic, cultural and religious implications. FAKEROM, with a project team reuniting a project leader, a young researcher, two postdoctoral students, and a MSc student, sets out to produce the first in-depth study of fake news in the Romanian language and a contrastive analysis with at least two other languages by combining qualitative instruments (on a linguistic base, originating primarily in critical discourse analysis) with quantitative tools (on a computational base, involving particularly natural language processing techniques). The originality of the FAKEROM project, a pioneering endeavor in the Romanian ethnolinguistic area (the Republic of Moldova included), resides in its interdisciplinary approach, which brings together humanities and computer science, the size of the surveyed corpus (roughly 12,000 texts), and the high granularity of the categories deployed to classify the different types of news reports. The project aims not only to generate significant scientific impact (both at a national and international level) but also to trigger a series of tangible economic, social, educational, and cultural effects outside its academic context.

SEMANTIC - Semantic Media Analytics Call name:
2020 - 2020

Role in this project: Project coordinator

Coordinating institution: UNIVERSITATEA POLITEHNICA DIN BUCURESTI

Project partners: UNIVERSITATEA POLITEHNICA DIN BUCURESTI ()

Affiliation: UNIVERSITATEA POLITEHNICA DIN BUCURESTI ()

Project website:

Abstract:

Ecosistem de cercetare, inovare și dezvoltare de produse și servicii TIC pentru o societate conectată la Internet of Things – NETIO

Realising an Applied Gaming Eco-system Call name:
H2020-ICT-2014/H2020-ICT-2014-1 2015 - 2019

Role in this project:

Coordinating institution: OPEN UNIVERSITEIT NEDERLAND

Project partners: OPEN UNIVERSITEIT NEDERLAND (); UNIVERSIDAD COMPLUTENSE DE MADRID (); INSTITUTO DE ENGENHARIADE SISTEMAS E COMPUTADORES (); PlayGen Ltd (); OKKAM SRL (); FORSCHUNGSINSTITUT FUR TELEKOMMUNIKATION UND KOOPERATION EV (); THE UNIVERSITY OF BOLTON (); TECHNISCHE UNIVERSITAET GRAZ (); INMARK ESTUDIOS Y ESTRATEGIAS SA (); UNIVERSITEIT UTRECHT (); INSTITUTO DO EMPREGO E FORMACAO PROFISSIONAL (); NUROGAMES GMBH (); BIP MEDIA SARL (); SOFIISKI UNIVERSITET SVETI KLIMENT OHRIDSKI (); STICHTING PRAKTIJKLEREN (); GAMEWARE EUROPE LIMITED (); Ministério da Justiça (); GROUPE RANDSTAD FRANCE (); HULL COLLEGE OF FURTHER EDUCATION (); UNIVERSITATEA POLITEHNICA DIN BUCURESTI ()

Affiliation: UNIVERSITATEA POLITEHNICA DIN BUCURESTI ()

Project website: http://rageproject.eu

Abstract:

RAGE was a flagship project combining the expertise of 19 partners from 10 countries for building assets aimed at helping with the development of serious games using state-of-the-art technologies and methods. It connected private companies, like game studios, with universities and public sector organizations.

Interpretarea Automata a imaginilor şi secvenţelor Video utilizând Procesarea Limbajului Natural Call name:
Cod SMIS 109513, contract 160/13.01.2017 2017 - 2019

Role in this project:

Coordinating institution: AUTONOMOUS SYSTEMS SRL

Project partners: AUTONOMOUS SYSTEMS SRL ()

Affiliation: AUTONOMOUS SYSTEMS SRL ()

Project website:

Abstract:

Hybrid automated machine integrating concurrent manufacturing processes, increasing the production volume of functional on-demand using high multi-material deposition rates Call name: EC - H2020
H2020-205448-723759 2016 - 2019

Role in this project:

Coordinating institution: FUNDACION AITIIP

Project partners: FUNDACION AITIIP (ES); VERO SOFTWARE LIMITED (UK); ALCHEMIE LIMITED (UK); CSEM CENTRE SUISSE D'ELECTRONIQUE ET DE MICROTECHNIQUE SA - RECHERCHE ET DEVELOPPEMENT (CH); CENTRO RICERCHE FIAT SCPA (IT); COMITE EUROPEEN DE COOPERATION DES INDUSTRIES DE LA MACHINE-OUTIL CECIMO AISBL (BE); AUTONOMOUS SYSTEMS SRL (RO); PININFARINA SPA (IT); PLANIT SOFTWARE LIMITED (UK); KARADIMAS DIMITRIOS (EL); WORLD PROFESSIONAL SERVICES SRLS.R.L (RO); ESPACE 2001 SA (LU); ARASOL ARAGONESA DE SOLDADURA SL (ES); ACCIONA CONSTRUCCION SA (ES); LEICA GEOSYSTEMS AG (CH); TWI LIMITED (UK); VERO UK LIMITED (GB)

Affiliation: AUTONOMOUS SYSTEMS SRL (RO)

Project website: http://www.krakenproject.eu/

Abstract:

Digital Preparing for Dialogues for Older-adult Consultations Call name:
18106 2018 - 2018

Role in this project:

Coordinating institution: University of Utrecht

Project partners: University of Utrecht (); University of Edinburgh (); DialogueTrainer B.V. (); UNIVERSITATEA POLITEHNICA DIN BUCURESTI (); Vilans ()

Affiliation: UNIVERSITATEA POLITEHNICA DIN BUCURESTI ()

Project website: http://prepdoc.eu

Abstract:

Empowering elderly people in conversations with healthcare professionals, Activity ID 18106. This project’s outcome was an application that prepares elderly people in having conversations with healthcare professionals through simulated communications. Patients’ free text answers were matched with the most semantically similar predefined answer using various NLP techniques.

Querying Databases in Natural Language Using Deep Learning Call name: P 2 - SP 2.1 - Proiect de transfer la operatorul economic
PN-III-P2-2.1-PTE-2016-0109 2016 - 2018

Role in this project:

Coordinating institution: BITDEFENDER S.R.L.

Project partners: BITDEFENDER S.R.L. (RO); UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)

Affiliation: UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)

Project website: https://rdprojects.bitdefender.com/proiecte-co-finantate-extern/pncdi-iii-2014-2020/text2neural

Abstract:

The Text2NeuralQL project primarily aims to build a solution for automatic generation of queries over structured databases (SQL) from texts written by users in natural language. To achieve our objective, we intend to use the latest advancements in natural language processing (NLP) and deep learning. We will build a comprehensive solution that encompasses both a model based on deep neural networks, especially recurrent networks, to generate SQL queries from texts, and an interactive dialog agent used to resolve more complicated queries, which can not be determined unambiguously from the entered text (because they are incomplete, ambiguous). For training and validating the models implemented in the project, we will build a dataset which will contain a set of natural language annotations of several data queries, labeled by experts/programmers. Two different datasets will be created: the former based on mining of information from online sources (communities of practice, discussion boards, question-answering sites) and the latter using crowdsourcing platforms, either already developed or built within the project. The developed solution addresses especially users from large companies without programming knowledge (marketing, sales managers) and startups who cannot afford to pay a programmer to query their databases for generating reports. Although several solutions for generating queries from texts have been constructed, this is a perfect time to develop a viable product given the advances achieved in recent years, in similar NLP tasks, including automatic translation and question-answering tasks. Recurrent deep models, such as LSTM or sequence-to-sequence, provided substantial improvements for many natural language tasks and we aim to expand them to generate qualitative queries from texts.

Smart and Networking UnderWAter Robots in Cooperation Meshes Call name: EC - H2020
H2020-197524-662107 2015 - 2018

Role in this project:

Coordinating institution: UNIVERSIDAD POLITECNICA DE MADRID

Project partners: UNIVERSIDAD POLITECNICA DE MADRID (ES); ECA ROBOTICS (FR); DEEPVISION AB (SE); INSTITUTO DE TELECOMUNICACOES (PT); INVENTAS AS (NO); EVOLOGICS GMBH (DE); ROBERT BOSCH GMBH (DE); HI IBERIA INGENIERIA Y PROYECTOS SL (ES); NEDERLANDSE ORGANISATIE VOOR TOEGEPAST NATUURWETENSCHAPPELIJK ONDERZOEK TNO (NL); STIFTELSEN SINTEF (NO); IXION INDUSTRY AND AEROSPACE SL (ES); TTI NORTE, S.L. (ES); FUNDACION TECNALIA RESEARCH & INNOVATION (ES); CONSORCIO PARA EL DISENO, CONSTRUCCION, EQUIPAMIENTO Y EXPLOTACION DE LA PLATAFORMA OCEANICA DE CANARIAS (ES); WATER LINKED AS (NO); AUTONOMOUS SYSTEMS SRL (RO); SINTEF AS (NO); SCUOLA SUPERIORE DI STUDI UNIVERSITARI E DI PERFEZIONAMENTO S ANNA (IT); SCIENCE AND TECHNOLOGY BV (NL); SABANCI UNIVERSITESI (TR); OFFICE NATIONAL D'ETUDES ET DE RECHERCHES AEROSPATIALES (FR); GREENSPHERE UNIPESSOAL LDA (PT); UNIVERSIDADE DE AVEIRO (PT); WORLD PROFESSIONAL SERVICES SRLS.R.L (RO); LEONARDO - SOCIETA PER AZIONI (IT); MAELARDALENS HOEGSKOLA (SE); DESISTEK ROBOTIK ELEKTRONIK YAZILIMAR-GE URETIM DANISMANLIK ITHALAT IHRACAT TICARET LIMITED SIRKETI (TR); THALES (FR); NORGES TEKNISK-NATURVITENSKAPELIGE UNIVERSITET NTNU (NO); ACCIONA CONSTRUCCION SA (ES); MARITIME ROBOTICS AS (NO)

Affiliation: AUTONOMOUS SYSTEMS SRL (RO)

Project website: http://www.swarms.eu

Abstract:

FILE DESCRIPTION	DOCUMENT
List of research grants as project coordinator or partner team leader
Significant R&D projects for enterprises, as project manager
R&D activities in enterprises
Peer-review activity for international programs/projects