/ Projects

We will develop a conceptual model of key components relating to supporting healthy behavior change. The model will provide a top-level representation of the clinical (from the psychological perspective) enablers and barriers that can be exploited for developing fine-grained models supporting the realization of behavior change paths within and across specific domains.

The resulting ontology will form the basis for generating user models (Theory of Mind), developing reasoning and decision-making strategies for managing conflicting values and motives, which can be used in collaborative and persuasive dialogues with the user. Such knowledge is also fundamental for embedding empathic behavior as well as non-verbal behaviors which can be embodied by a virtual character in the role of a coach. Learning methods can be applied to explore trajectories of behavior change. The produced ontology will represent a valuable resource for the healthcare domain thanks to the knowledge included into the provided resource.

Output

1 conference paper containing the description of the ontology and guidelines for its usage

1 ontology artifact

Presentations

Project Partners:

  • Fondazione Bruno Kessler (FBK), Mauro Dragoni
  • Centre national de la recherche scientifique (CNRS), Jean-Claude Martin
  • Umeå University (UMU), Helena Lindgren

Primary Contact: Mauro Dragoni, FBK

Main results of micro project:

This micro-project led to three tangible outputs:
– 1 conference paper presented at the AI4Function workshop of IJCAI titled "An Ontology-Based Coaching Solution for Increasing Self-Awareness of Own Functional Status".
– 1 journal paper currently under revision at the Data Intelligence Journal (MIT Press) titled "Modeling a Functional Status Knowledge Graph
For Personal Health".
– 3 extensions of the HeLiS ontology publicly released and accessible at https://horus-ai.fbk.eu/helis/
Concerning the last point, the work done in this microproject worked as trigger for the creation of a new knowledge graph of functional status.

Contribution to the objectives of HumaneAI-net WPs

The resulting ontology will form the basis for generating user models (Theory of Mind), developing reasoning and decision-making strategies for managing conflicting values and motives, which can be used in collaborative and persuasive dialogues with the user. Such knowledge is also fundamental for embedding empathic behavior as well as corresponding non-verbal behaviors which can be displayed by a virtual character embodying the role of a coach or the dispatching of motivational human-computer interactions over different devices (e.g. mobile phone and smartwatch).
The micro-project relates to WP6 (T6.5 and AI coach for behavior change), WP3 (T3.4 User models and interaction history, and empathy in human-AI systems) and WP5 (explanation of medical advice, conflicting values and moral limits of nudging).

Tangible outputs

Attachments

3.11 The Knowledgeable Coach FBK CNRS UMU pitch-video_Berlin.mkv

With the rise of social media platforms, we have witnessed the emergence of alarming phenomena in public debates, such as the polarization of opinions. The Friedkin-Johnsen model is a very popular model in opinion dynamics, validated on real groups, and well-investigated from the opinion polarization standpoint. Previous research has focused almost exclusively on static networks, where links between nodes do not evolve over time. However, it has been shown that they can break down if they are not strong enough to sustain disagreement.

In this micro-project, we want to fill this gap by designing a variant of the Friedkin-Johnsen model that embeds the dynamicity of social networks. Furthermore, we will design a novel definition of global polarization that combines network features and opinion distribution, to capture the existence of clustered opinions. We will analyse the polarization effect of the new dynamic model, and identify the impact of the network structure.

Output

Code of the simulator for the new dynamic model

Working paper, to be published in a workshop (possibly conference)

Project Partners:

  • Consiglio Nazionale delle Ricerche (CNR), Elisabetta Biondi
  • Central European University (CEU), Janos Kertesz
  • Central European University (CEU), Gerardo Iniguez

 

Primary Contact: Elisabetta Biondi, CNR-IIT

Results Description

Human social networks are very complex systems and their structure has an essential impact on opinion dynamics. However, since my main goal is to study the impact of the opinion dynamics model per se, we decided to deal with two different social network typologies: a Erdős–Rényi (ER) and a stochastic block model (SBM).

Design of the FJ dynamic model.We have implemented a rewiring policy that has been extensively studied in discrete opinion diffusion models. This involves substituting edges that connect nodes with different opinions with other edges. We have adapted this scheme to work with the FJ model's opinions, which are within the range of [-1,1], in both the asynchronous and synchronous versions. According to two parameters θ (the disagreement threshold) and p_rew (the rewiring probability):
– With probability 1-p_rew the FJ is applied
– With probability p_rew, if i and j disagree, i.e. |x_i-x_j |> θ, the edge (i,j) is replaced with an edge (i,k) where k agrees with i, i.e. |x_i-x_j |<= θ.
The above algorithm was specifically designed and implemented for the ER graph. However, in the case of the SBM, I have limited the potential candidates for rewiring to nodes within a maximum of two hops distance. This decision was made to prevent the block structure from becoming entirely irrelevant. The rationale behind this choice is based on the triadic closure mechanism, which suggests that individuals are more inclined to choose new acquaintances among the friends of their friends.

Design of the polarization metric.The design of the polarization metric involved developing a definition for identifying highly polarized networks. We defined a highly polarized network as one in which there are two distinct opinions that are clustered into two tightly connected communities. To achieve this, we needed to consider both the network structure and the distribution of opinions. Therefore, we decided to use two different metrics to measure these aspects: bimodality for the opinion distribution and homogeneity for its correspondence with the network structure.

Bimodality.The bimodality coefficient was used to measure the extent to which a distribution is bimodal. It is calculated using the skewness and kurtosis values and represents how similar the distribution is to one with two modes.

Homogeneity.To measure the homogeneity of the opinion distribution with the network structure, we examined the local distribution of nodes' opinions. We looked at whether each node's opinion was similar to those of its neighbors, which would suggest that it was in line with the overall opinion distribution over the network. The final homogeneity value was close to zero if the distribution of opinions was close to linear.

Experimental evaluation.We have developed a Python simulator that can compute the dynamic FJ (rewiring included), and polarization metrics over time based on the given network and initial opinions. To test the model, we ran simulations on a small network comprising 20 nodes and compared the outcomes of the FJ with rewiring to those without rewiring. For the ER network, we used a vector of uniformly distributed opinions over [-1,1] as the initial opinions. However, for the SBM networks, we employed a different configuration, where the initial opinions were uniformly extracted over the intervals [-0.5,0-0.1] and [0.1,0.5], depending on whether the nodes belonged to one or the other block.

In conclusion, this microproject involves the design of a dynamic version of the FJ model for synchronous and asynchronous cases. Additionally, we have developed a new definition of polarization that considers both the distribution of opinions and the network topology. To assess the model's effectiveness, we conducted simulations on two different network types: an ER network and an SBM network. Our findings indicate that the rewiring process has significant effects on polarization, but these effects are dependent on the initial network.

Publications

No publications yet. The collaboration is still ongoing.

Links to Tangible results

Github link of the code of the simulator for the new dynamic model:
https://github.com/elisabettabiondi/FJ_rewiring_basic.git

Collective intelligence and a deliberative democratic system rest on the shoulders of a public free access to unbiased and diverse information. Social media has become a source of news and an important factor in opinion-formation. However, social media is not a neutral terrain but can be a platform for political manipulation.

This project aims to understand the cognitive, socio-affective, and neurobiological factors impacting our judgement of and response to information.This can conclude to behavioural recommendations that empower us to protect ourselves from manipulation and disinformation, and a targeted design of interventions to prevent their spread and the effective allocation of resources to protect those most vulnerable.

Output

Paper (aiming for Communication, Nature Human Behavior or similar)

Project Partners:

  • ETH Zurich (ETHZ), Elisabeth Stockinger
  • Fondazione Bruno Kessler (FBK), Riccardo Gallotti

Primary Contact: Elisabeth Stockinger, ETH Zurich

Results Description

Despite all efforts to mitigate mis- and disinformation, they continue to be a substantial problem. This project contributed to the literature base on mis- and disinformation about social media with an analysis of the interaction effects between temporal rhythms of disinformation and social media usage in the context of the COVID-19 pandemic.
Specifically, consider how mis- and disinformation spread on Twitter varies throughout the day and whether there are individual differences in users' propensity to spread mis- and disinformation on Twitter based on the activity patterns.

We analysed a comprehensive dataset, examining the reliability of information relating the COVID-19 pandemic shared on Twitter. We clustered users into pseudo-chronotypes based on their activity patterns on Twitter throughout the day, identified times of waking and prolonged waking states per cluster as well as times of increased susceptibility.

We aggregated out results into a paper and submitted an extended abstract to the International Conference on Computational Social Science (www.ic2s2.org/submit_abstract), acceptance pending, and are in the process of preparing a paper for submission for a reputable journal.

Elisabeth Stockinger from ETHZ spent a 3-week mobility period at FBK in Trento, Italy, to work with the partner directly. She has continued the collaboration as a virtual visiting student.

Publications

We submitted an extended abstract to the International Conference on Computational Social Science (www.ic2s2.org), acceptance pending.

Links to Tangible results

None yet.

Current speech translation data sets contain pre-segmented speech audio, post-processed transcripts, and reference translations. Such data do not allow identifying error contributions of individual components in the whole speech translation pipeline and often lack detailedness to identify major error contributors. There is also lack of data for evaluating speech translation of web-based meetings (Zoom, Microsoft Teams, etc.) that have become vital to enable remote work and remote international cooperation.

To address the gaps, we propose:

1. to collect a data set of multi-speaker covering various domains and languages.

2. to create multi-layer annotations of the data that would allow evaluating individual components of pipeline-based (and also end-to-end) speech translation systems and measure each component's contribution towards the total error.

The data will feature audio data from multiple languages (including Latvian and Lithuanian for which no speech translation evaluation sets have ever been created).

Output

Data set for evaluation of speech translation systems consisting of audio data (in at least 3 languages)

Multi-layer annotations of the data (with at least the following annotation layers – speaker segmentation, sentence segmentation, orthographic transcriptions, normalised transcriptions, translations), and documentation of the data.

Project Partners:

  • Charles University Prague, Ondrej Bojar

Primary Contact: Aivars Bērziņš, Tilde

This second micro project of WP6.10 will validate the air quality prototype developed in the first micro project with a second, real city, Valladolid in Spain. In principle, there will be no new developments except for feedback about the system from the city. This project is also in line with the objective of Humane AI: to shape the AI revolution in a direction that is beneficial to humans both individually and societally, and that adheres to European ethical values and social, cultural, legal, and political norms. The focus of the project will be on insights generated from data collected with mobile air quality measurement stations to be placed on top of vehicles, and to test how does insides help local governments in better managing the challenges around air quality.

Output

A report with the results of the evaluation of the prototype.

Adaptations to the system based on feedback from the local city.

Dissemination activities jointly by Telefonica and the local city.

Potential policy measures to improve the air quality in Valladolid.

Project Partners:

  • Telefónica Investigación y desarrollo S.A. (TID), Richard Benjamins

Primary Contact: Richard Benjamins, Telefonica (TID)

This micro-project aims to explore the integration of virtual coaching and the use of Automated Dietary Monitoring wearable devices by conducting a preliminary study in which these technologies are integrated and experimented in a real-life context. Such a capability will enable the investigation about the role of real-time sensing data within a healthcare monitoring system supporting users by suggesting the most appropriate healthy behavior. The designed strategy will be validated within a living lab involving a group of 20-30 users that will wear the textile and will use the application for a period of three weeks.

Goals are to validate the capability of the smart textile in detecting chewing activities in a real-world environment, the effectiveness of the coaching system in detecting unhealthy dietary behaviors, the quality of the feedback provided, and the overall acceptability of the users with respect to a coaching system introducing a minimum invasive strategy.

Output

1 conference paper containing the description of the proposed approach together with the results and the insights gathered from the living lag we run.

1 dataset containing all the generated data that will be made available to the HumanE-AI network

Project Partners:

  • German Research Centre for Artificial Intelligence (DFKI), Paul Lukowicz
  • Fondazione Bruno Kessler (FBK), Mauro Dragoni

Primary Contact: Mauro Dragoni, FBK

The basic challenge regarding debiasing ML models is that in order to prevent models from generating bias on the basis of some sensitive characteristics it is necessary to have information about these characteristics. Usually this information is not available Fortunately, there is a new approach: Adversarial Reweighted Learning which debiases the models without having sensitive attribute information. However this approach redefines the fairness as Rawlsian max-min principle which is quite different from parity based fairness definitions that have been hitherto used. The goal of this project is to scrutinize the implications of using Rawlsian fairness principle in order debias the models by scrutinizing three things

1. Are there sensitive attributes for which Rawlsian fairness is unsuitable?

2. What would be the parity-based fairness scores when Rawlsian fairness definition is used for debiasing the models?

3. Is it possible to use Rawlsian fairness (and ARL) as post-processing method to existing models?

Output

Publication on identifying the effect of Rawlsian fairness on parity-based fairness definitions

Publication on how ARL can be used for models that are already being used as a post-processing tool

Project Partners:

  • ING Groep NV, Dilhan Thilakarathne

Primary Contact: Dilhan Thilakarathne, ING

Blockchains such as Bitcoin and Ethereum currently are computational wasteful. On an annual basis both blockchains consume over a 70 terawatt-hours (TWh) of energy on computational resources to secure the network, which is similar to the annual energy consumption of Switzerland. These computational resources are used to reverse a cryptographic hash function (which is called a consensus algorithm) that is a solution to a puzzle, but serves no other purpose. Such amount of computational resources should be used more efficiently. Our aim is to use the large amount of computational resources more efficient by replacing the cryptographic hash function with a machine learning task. We focus on the Ethereum network as its computational power is not limited to solving hash functions only, as is mostly the case in Bitcoin. However, our intended solution can be generalized to any blockchain that is currently a computational wasteful.

Output

Publication: Review paper

Project Partners:

  • ING Groep NV, Dilhan Thilakarathne

Primary Contact: Dilhan Thilakarathne, ING

Algebraic Machine Learning (AML) offers new opportunities in terms of transparency and control. However, that comes along with many challenges regarding software and hardware implementations. To understand the hardware needs of this new method it is essential to analyze the algorithm and its computational complexity. With this understanding, the final goal of this microproject is to investigate the feasibility of various hardware options particularly in-memory processing hardware acceleration for AML.

Output

Simulation model for a PIM architecture using AML

Report

Presentations

Project Partners:

  • Algebraic AI S.L., Fernando Martin Maroto
  • Technische Universität Kaiserslautern (TUK), Christian Weis
  • German Research Centre for Artificial Intelligence (DFKI), Matthias Tschöpe

Primary Contact: FERNANDO MARTIN MAROTO, Algebraic AI

Main results of micro project:

We have carried out a theoretical study of the AML sparse crossing algorithm efficiency and identified in-memory processing and FPGA combined with in-memory processing as the two feasible options for Algebraic Machine Learning. Currently, we are working on a prototype implementation that involves FPGA and in-memory processing of bit arrays in commercial Upmem RAM memories.

Contribution to the objectives of HumaneAI-net WPs

This work is critical to speed up the calculation of Algebraic Machine Learning models and in so doing contribute to:
1- Bidirectional human-machine communication using formal expressions
2- Possibility to set goals and establish limits via formal constraints
3- Reduced dependency on statistics can help overcome bias
4- Transparency by design
5 -Possibility for decentralized, cooperative distributed machine learning

Tangible outputs

  • Program/code: AML engine prototype using bitarrays – Fernando Martin Maroto
    www.algebraic.ai

Results Description

Sparse Crossing is a machine learning algorithm based on algebraic semantic embeddings. The goal of the collaboration is to first understand the needs and computational complexity of Sparse Crossing and then perform a feasibility analysis of various hardware options for an efficient implementation of the algorithm. Particularly, in-memory processing hardware acceleration and FPGA-based implementations have been considered.

A report, and a FPGA based prototype has been developed.

Publications

Fernando Martin-Maroto and Gonzalo G. de Polavieja. Algebraic Machine Learning.
arXiv:1803.05252, 2018.
Fernando Martin-Maroto and Gonzalo G. de Polavieja. Finite Atomized Semilattices.
arXiv:2102.08050, 2021.
Fernando Martin-Maroto and Gonzalo G. de Polavieja. Semantic Embeddings in Semilattices. arXiv:2205.12618, 2022

Links to Tangible results

The report and the details of the FPGA based prototype are available for review. Some results are under a patent process and are not yet available to the public.

This micro project will study the adaptation of automatic speech recognition (ASR) systems for impaired speech. Specifically, the micro-project will focus on improving ASR systems for speech from subjects with dysarthria and/or stuttering speech impairment types of various degrees. The work will be developed using either German “Lautarchive” data comprising only 130 hours of untranscribed doctor-patient German speech conversations and/or using English TORGO dataset. Applying human-in-the-loop methods we will spot individual errors and regions of low certainty in ASR in order to apply human-originated improvement and clarification in AI decision processes.

Output

Paper for ICASSP 2021 and/or Interspeech 2022

Presentations

Project Partners:

  • Brno U, Mireia Diez
  • Technische Universität Berlin (TUB), Tim Polzehl

Primary Contact: Mireia Diez Sanchez, Brno University of Technology

Main results of micro project:

Project has run for less than 50% of its allocated time.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Contribution to the objectives of HumaneAI-net WPs

WP1 Learning, Reasoning and Planning with Human in the Loop
T1.1 Linking symbolic and subsymbolic learning

WP3 Human AI Interaction and Collaboration
T3.1 Foundations of Human-AI interaction and Collaboration
T3.6 Language-based and multilingual interaction
T3.7 Conversational, Collaborative AI

WP6 Applied research with industrial and societal use cases
T6.3 Software platforms and frameworks
T6.5 Health related research agenda and industrial usecases

Tangible outputs

  • Publication: –
  • Other: Internal report – Mireia Diez Sanchez, mireia@fit.vutbr.cz

Results Description

In this miscroproject, we pursued enabling access to AI technology to those who might have special needs when interacting with AI: Automatic Speech Recognition made accessible for people with dysarthria

Dysarthria is a motor speech disorder resulting from neurological injury and is characterized by poor articulation of phonemes. Within Automatic speech recognition (ASR), dysarthric speech recognition is a tedious task due to the lack of supervised data and diversity.

Particularly, in this work, we have studied the performance of different ASR systems on dysarthric speech: LF-MMI, Transformer and wav2vec2. The analysis revealed the superiority of the wav2vec2 models on the task. We investigated the importance of speaker dependent auxiliary features such as fMLLR and xvectors for adapting wav2vec2 models for improving dysarthric speech recognition. We showed that in contrast to hybrid systems, wav2vec2 did not improve by adapting model parameters based on each speaker.
We proposed a wav2vec2 adapter module that inherits speaker features as auxiliary information to perform effective speaker normalization during finetuning. We showed that, using the adapter module, fMLLR and xvectors are complementary to each other, and proved the effectiveness of the approach outperforming existing SoTA on UASpeech dysartric speech ASR.
In our cross-lingual experiments, we also showed that combining English and German data for training, can further improve performance of our systems, proving useful in scenarios where little training examples exist for a particular language.

AdAIS addressed topics related to the tasks:
WP1 Learning, Reasoning and Planning with Human in the Loop
T1.1 Linking symbolic and subsymbolic learning
WP3 Human AI Interaction and Collaboration
T3.1 Foundations of Human-AI interaction and Collaboration
T3.6 Language-based and multilingual interaction
T3.7 Conversational, Collaborative AI
WP6 Applied research with industrial and societal use cases
T6.3 Software platforms and frameworks
T6.5 Health related research agenda and industrial use cases

Publications

Publication: M. K. Baskar, T. Herzig, D. Nguyen, M. Diez, T. Polzehl, L. Burget, J. Černocký, “Speaker adaptation for Wav2vec2 based dysarthric ASR”. Proc. Interspeech 2022, 3403-3407, doi: 10.21437/Interspeech.2022-10896

Links to Tangible results

Link to publication: https://www.isca-speech.org/archive/pdfs/interspeech_2022/baskar22b_interspeech.pdf

Open source tool for training ASR models for dysarthic speech: https://github.com/creatorscan/Dysarthric-ASR
The repository contains: A baseline recipe to train a TDNN-CNN hybrid model based ASR system, this recipe is prepared to be trained on the TORGO dataset. And an end-to-end model using ESPnet framework prepared to be trained on UASpeech dataset.

Understanding the mechanism of the neural correlates during human physical activities is important for providing safety in industrial factory environments considering brain activity during lifting a weight. Moreover, different responses to the same task can be observed due to physiological and neurological differences among individuals. In this project, the change pattern in EEG will be investigated during lifting of a weight and the features in EEG data making difference during lifting a weight will be analyzed. Classification between lifting and no lifting cases will be realized by using deep learning based machine learning methods. The outcomes of the project can be applied in industrial exoskeleton applications as well as physical rehabilitation of stroke patients.

Output

Dataset Repository (Share on AI4EU)

Conference Paper / Journal Article

Presentations

Project Partners:

  • Türkiye Bilimsel ve Teknolojik Araştırma Kurumu (TUBITAK), Sencer Melih Deniz
  • German Research Centre for Artificial Intelligence (DFKI), Paul Lukowicz

Primary Contact: Sencer Melih Deniz, TUBITAK BILGEM

Main results of micro project:

The project has run for almost 50% of its allocated time and has yet to be completed. Within this time duration, the following steps were completed:
1. Experimental paradigm was designed to achieve the project goals.
2. Study preparation including hardware and software development was completed.
3. Data recording session has been started and is in progress. Data from a total of 10 people has been obtained so far. More participants will be included in data acquisition to achieve the desired result.

The dataset and results will be evaluated once the data acquisition is completed.

Contribution to the objectives of HumaneAI-net WPs

This project is also part of WP2 with task numbers T2.2, T2.3.

This project aims to contribute to WP2 and WP6 by investigating the use case of EEG signal and AI models in the detection of various aspects of physical activities during weightlifting. To investigate pattern change in EEG during weightlifting will be aimed at providing more information in prediction of intended and actual human actions during sensori-motor tasks. Doing so, a common research question is aimed to be applied to the more industrial use cases such as control of exoskeletons. Moreover, outcomes of the project can be used for contribution in increasing mobility in stroke patients and disabled people as related with healthy living and mobility.

Tangible outputs

  • Program/code: Data Acquisition Software Code – Juan Felipe Vargas Colorado

Attachments

PresentMovie_NeuralMech_WLlifting_TUBITAK_DFK_Berlin.m4v

Results Description

In this project, it was investigated whether EEG (electroencephalography) signal can be used for detecting the motion as well as the variable weights a person is lifting. To do this, an experimental paradigm has been designed and EEG data have been acquired during performing biceps flexion-extension motions for different weight categories: lifting with no weight (empty), medium, and heavy lifting.

Features in EEG data generating difference for each lifted weight of category have been investigated. EEG data via different two EEG headsets have been collected from various participants while they lift different categories of load, namely empty, medium and heavy, in this project. Then, EEG data have been analyzed to realize if different category of weigths result in difference in EEG data by applying different deep learning methods together with different machine learning methods. According to the obtained results, it can be said that that EEG signals can be successfully used as a method to predict different loads during dynamic bicep curl motion. Therefore, this result could result more researches to develop rehabilitation systems robust to dynamic changes in weight. Moreover, information regarding weight change could contribute to a better estimation of fatigue condition to be used in sports and training applications. Finally, it has been evaluated that the approach to predict different categories of lifted weight could be used in further optimizations in industrial applications for which usage of exoskeleton can be given as an example.

Results of micro project in which TUBITAK BILGEM and Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI) Kaiserslautern collaborated was presented at the IEEE-EMBS International Conference on Biomedical and Health Informatics jointly organised with the IEEE-EMBS International conference on Wearable and Implantable Body Sensor Networks organized in Ioannina, Greece between 27-30 September 2022. Also it was published with the title “Prediction of Lifted Weight Category Using EEG Equipped Headgear" in 2022 IEEE-EMBS International Conference on Biomedical and Health Informatics Conference Proceedings.

Publications

Conference proceeding:
"Prediction of Lifted Weight Category Using EEG Equipped Headgear", published in 2022 IEEE-EMBS International Conference on Biomedical and Health Informatics.

Links to Tangible results

Paper: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9926744
Dataset: https://www.ai4europe.eu/research/research-bundles/neural-mechanism-human-brain-activity-during-weight-lifting?category=ai_assets

Knowledge discovery offer numerous challenges and opportunities. In the last decade, a significant number of applications have emerged relying on evidence from the scientific literature. ΑΙ methods offer innovative ways of applying knowledge discovery methods in the scientific literature facilitating automated reasoning, discovery and decision making on data.

This micro-project will focus on the task of question answering (QA) for the biomedical domain. Our starting point is a neural QA engine developed by ILSP addressing experts’ natural language questions by jointly applying document retrieval and snippet extraction on a large collection of PUBMED articles, thus, facilitating medical experts in their work. DFKI will augment this system with a knowledge graph integrating the output of document analysis and segmentation modules. The knowledge graph will be incorporated in the QA system and used for exact answers and more efficient Human-AI interactions. We will primarily focus upon scientific articles on Covid-19 and SARS-CoV-2.

Output

Paper(s) in a conference or/and journal

Demonstrator

Presentations

Project Partners:

  • ATHINA, Haris Papageorgiou
  • German Research Centre for Artificial Intelligence (DFKI), Georg Rehm

Primary Contact: Haris Papageorgiou, ATHENA RC/Institute for Language & Speech Processing

Results Description

Knowledge discovery offer numerous challenges and opportunities. In the last decade, a significant number of applications have emerged relying on evidence from the scientific literature. ΑΙ methods offer innovative ways of applying knowledge discovery methods in the scientific literature facilitating automated reasoning, discovery and decision making on data.

This micro-project focused on the task of question answering (QA) for the biomedical domain. ILSP developed a neurosymbolic QA engine addressing experts’ natural language questions by jointly applying neural snippet retrieval, biomedical Knowledge Graphs and our previous work on neural QA on a large collection of PUBMED articles, thus, facilitating medical experts in their work. DFKI further augmented this system by integrating the output of document analysis and segmentation modules. The final QA system supports exact answers, a diverse set of questions and more efficient Human-AI interactions. Moreover, a demonstrator was built upon scientific articles on Covid-19 and SARS-CoV-2.

Publications

Pappas Dimitris, Lyris Ioannis, Kountouris George and Papageorgiou Haris: "A Neurosymbolic Question Answering System Combining Structured and Unstructured Biomedical Knowledge", Proceedings of the 3rd Conference on AI for Humanity and Society (AI4HS), Stockholm, 2022

Links to Tangible results

Video: "Combining symbolic and sub-symbolic approaches – Improving neural Question-Answering-Systems through Document Analysis for enhanced accuracy and efficiency in Human-AI interaction"

Demonstrator: A neurosymbolic Question Answering System on Covid-19 and SARS-CoV-2

Humane AI on Social Media

EU Logo

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 952026.

Terms of Service & Privacy Policy | © Knowledge 4 All Foundation