We congratulate Georg Wenzel who has received the best master thesis award at the Department of Computer Science for 2024 with the prize of 500 eur. Georg's work [1] is on using context to predict changes in validity duration of text statements and, by this, to be able to better assess information obsoleteness.

[1] Georg Wenzel, Adam Jatowt: Temporal Validity Change Prediction. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024), ACL Press, findings paper (2024) PDF

A new PhD student joins the DS lab. Mohammed Mustafa, who was a lecturer at Assiut University, will start working on supporting financial report writing according to the new EU taxonomy regulations in the framework of the new FFG project: “AI Enabled Sustainability Jurisdiction Demonstrator”. Mohammed has previously worked on aspect-based sentiment mining.

We are happy to host Ulin Nuha from the National Kaohsiung University of Science and Technology, Taiwan for a 3-months' long research stay at the DS group. Ulin is a PhD student and will work on machine translation from Ladin to Italian and English. His research stay is financed from the Ernst Mach Grant by OeAD.

The paper [1] about the method for automatic keyword extraction from text co-authored by Prof. Adam Jatowt has become the top most highly cited paper of the last 5 years in the Information Sciences journal (Elsevier) [IF: 8.23].

[1] R Campos, V Mangaravite, A Pasquali, A Jorge, C Nunes, A Jatowt: YAKE! Keyword extraction from single documents using multiple local features Information Sciences 509, 257-289 (2020)

A paper on a holistic approach to multiple-choice visual question answering based on generating questions, answers and distractors [1] has been accepted to appear at EMNLP 2024 conference (A* conference). Additionally a paper introducing a novel method for context preparation based on automatic hints for improving question answering by LLMs [2] as well as a short paper on detecting temporally ambiguous questions [3] have been accepted to appear in the findings of EMNLP 2024.

[1] Wenjian Ding, Yao Zhang, Jun Wang, Adam Jatowt, Zhenglu Yang: Exploring Union and Intersection of Visual Regions for Generating Questions, Answers, and Distractors. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024), ACL Press, (2024)

[2] Jamshid Mozafari, Abdelrahman Abdallah, Bhawna Piryani, Adam Jatowt: HiGenQA: Exploring Hint Generation Approaches for Open Domain Question Answering. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024), ACL Press, findings paper, (2024)

[3] Bhawna Piryani, Abdelrahman Abdallah, Jamshid Mozafari, Adam Jatowt: Detecting Temporal Ambiguity in Questions. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024), ACL Press, findings paper, (2024)

The Data Science group has received an FFG project on the use of LLMs for supporting company report writing. The “AI Enabled Sustainability Jurisdiction Demonstrator” FFG-sponsored project brings together 12 partner institutions from both academia and industry to research and develop new technologies for automating EU taxonomy based project writing. The goal is to increase the number of companies in Austria that can independently meet the requirements of sustainability reporting in relation to the EU Taxonomy through large language models involvement. The project under the call of “Digitale Technologien 2023 - Kooperationsprojekte und Sondierungen” will last from 2024 to 2027.

A new book authored by Vaibhav Mavi (New York University), Anubhav Jangra (Columbia University) and Adam Jatowt (University of Innsbruck) on the topic of multi-hop question answering has been published [1]. Multi-hop Question Answering involves answering complex questions that require several hops (e.g., accessing information from several documents) in order to be answered. This necessitates the combination of effective information retrieval, natural language processing and inference techniques.

[1] Vaibhav Mavi, Anubhav Jangra, Adam Jatowt: Multi-hop Question Answering, Foundations and Trends in Information Retrieval (FnT Information Retrieval), Book, pp 457-586. http://dx.doi.org/10.1561/1500000102 (2024)

The DS lab has got 1 Full, 3 Resource and 1 Short paper accepted at SIGIR 2024 conference - the main event for disseminating Information Retrieval research and an A* conference which was held in July, 2024 in Washington, DC, USA. The accepted papers are listed in the publications section.

A findings full paper [1] based on the master thesis of Georg Wenzel appeared at the ACL 2024 conference (A* conference) as a findings paper. The paper is about assessing the impact of context on measuring the expected validity period of information, which can be useful in detecting obsolete text content. Another paper about intelligent example selection for in-context learning has appeared as a short findings paper [2].

[1] Georg Wenzel, Adam Jatowt: Temporal Validity Change Prediction. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024), ACL Press, findings paper (2024) PDF

[2] Yiquan Wu, Anlai Zhou, Yuhang Liu, Yifei Liu, Adam Jatowt, Weiming Lu, Jun Xiao, Kun Kuang: Chain-of-Quizzes: Pedagogy-inspired Example Selection in In-Context-Learning. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024), ACL Press, findings paper, pp. 10136–10142 (2024) PDF

We are happy to announce that the Seventh International Workshop on Narrative Extraction from Texts (Text2Story 2024) held in conjunction with the 46th European Conference on Information Retrieval on March 24, 2024 in Glasgow will host 17 presentations including 2 keynotes. More information about the workshop's program is available at is homepage.

We have just concluded the publication of a special issue at the International Journal on Digital Libraries (IJDL, Springer) which gathers the best papers from the 24th International Conference on Asia-Pacific Digital Libraries (ICADL 2022).

We are happy to host Yiquan Wu from Zhejiang University, China for a 6-months' long research stay at DS group. Yiquan is a PhD student at Zhejiang University and he has been doing internship at Alibaba since 2018. He works on question answering/generation in the legal domain pushing the boundaries of LegalAI.

The paper co-authored by an UIBK student Lukas Eder on contrastive keyword extraction [1] has become a Runner-up for the Best Demo Paper Award at the 32nd ACM International Conference on Information and Knowledge Management (CIKM 2023).

[1] Lukas Eder, Ricardo Campos and Adam Jatowt: Contrastive Keyword Extraction from Versioned Documents. Proceedings of the 32nd ACM International Conference on Information and Knowledge Management (CIKM 2023), ACM Press, pp. 5026-5030 (2023).

Adam Jatowt has gave a keynote at the Large Language Models' Interpretation and Trustworthiness (LLMIT) workshop organized in conjunction with CIKM 2023 conference. The talk's title was “Temporal Reasoning Capabilities of Large Language Models”.

A paper on analyzing the level of temporal factual knowledge of Large Language Models and their temporal blind spots have been accepted at the 17th ACM International WSDM Conference [1]. This work has been done in collaboration with Jonas Wallat from L3S research center from and Avishek Anand from Delft University.

[1] Jonas Wallat, Adam Jatowt, Avishek Anand: Temporal Blind Spots in Large Language Models. Proceedings of the 17th ACM International WSDM Conference (WSDM 2024), ACM Press (to appear in 2024)

A paper on analyzing the ability of large language model on temporal commonsense reasoning [1], and a paper introducing a dataset for text classification, text generation, text summarization and question answering in legal domain [2] have been accepted to appear at EMNLP 2023 conference (Core A* conference).

[1] Raghav Jain, Daivik Sojitra, Arkadeep Acharya, Sriparna Saha, Adam Jatowt, Sandipan Dandapat: Do Language Models Have a Common Sense regarding Time? Revisiting Temporal Commonsense Reasoning in the Era of Large Language Models, Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP 2023) ACL Press, (2023r)

[2] Changzhen Ji, Yating Zhang, Adam Jatowt and Haipang Wu: CDD: A Large Scale Dataset for Legal Intelligence Research, Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP 2023) ACL Press, (2023)

Prof. Jatowt gave a talk on Large Language Models at “AI in Tourism: Relevance, Opportunities and Risks” symposium organized in Innsbruck (10 October 2023) and another talk at “Generative AI: Use and implementation of LLMs” workshop organized by the City Agency of Innsbruck (12 October 2023).

The bachelor's thesis work of Lukas Eder on contrastive keyword extraction [1] has been accepted at the 32nd ACM International Conference on Information and Knowledge Management (CIKM 2023).

[1] Lukas Eder, Ricardo Campos and Adam Jatowt: Contrastive Keyword Extraction from Versioned Documents. Proceedings of the 32nd ACM International Conference on Information and Knowledge Management (CIKM 2023), ACM Press, pp. 5026-5030 (2023).

Maxime Guillemaud who is a master student at the Aix-Marseille University stays with us from 15 May, 2023 for three months as a visiting research student. Maxime is working on topics related to humor detection and analysis in text.

We are happy to invite Anubhav Jangra from Google India for a short visit to our group. Anubhav works among others on topics of multimodal document summarization and advertisement slogan generation.

A paper on temporally-enhanced language model [1] and a paper about datasets for humor analysis [2] have been accepted to appear at SIGIR 2023 conference (Core A* conference).

[1] Jiexin Wang, Adam Jatowt, Masatoshi Yoshikawa and Yi Cai: Extending Pre-Trained Language Representations with Dual-Type Temporal Information. Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2023), ACM (2023)

[2] Liana Ermakova, Anne-Gwenn Bosser, Adam Jatowt and Tristan Miller: The JOKER Corpus: English–French Parallel Data for Multilingual Wordplay Recognition. Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2023), ACM (2023)

Prof. Adam Jatowt and his two collaborators Taishi Hosokawa and Kazunari Sugiyama have won the best paper award at ECIR 2023, the 45th European Conference on Information Retrieval, which is core A ranked conference, for the paper on temporal text validity [1].

[1] Taishi Hosokawa, Adam Jatowt and Kazunari Sugiyama: Temporal Natural Language Inference: Evidence-based Evaluation of Temporal Text Validity. Proceedings of the 45th European Conference on Information Retrieval (ECIR 2023), Dublin, Ireland, Springer LNCS, pp. 441-458 (2023) PDF

We are happy to have Assoc. Prof. Yu Suzuki from Gifu University, Japan visiting our research group from March 10 to March 31, 2023. Prof. Yu Suzuki does research on social media mining, and Wikipedia content analysis as well as processing.

Jamshid Mozafari has joined DS group as a PhD student starting from March 2023. Jamshid has carried in the past research on automatic question answering (QA) including Persian language QA.

Our study on fine-grained sentiment analysis of tweets in San Francisco and London [1] has been reported in Ö1-Wissenschaft (science.orf.at), Bloomberg news, Yahoo news, International Business Times and 16 other media outlets.

[1] Panote Siriaraya, Yihong Zhang, Yukiko Kawai, Péter Jeszenszky, Adam Jatowt: A City-Wide Examination of Fine-Grained Human Emotions through Social Media Analysis, PloS One, 18(2), (2023) PDF

Adam Jatowt has been granted the prestigious International Friedrich Wilhelm Bessel Research Award by Humboldt Society. The award bears the name of the German astronomer and mathematician Friedrich Wilhelm Bessel (1784-1846), and is funded by the German Federal Ministry of Education and Research.

Humour remains one of the most difficult aspects of intercultural communication. Understanding humor often involves understanding implicit cultural references and/or double meanings, especially in the case of wordplay, which raises not only the question of its (un)translatability, but also how to detect and classify instances of this complex phenomenon. The goal of the JOKER track is to bring together linguists, translators, and computer scientists to create reusable test collections for benchmarking and to explore new evaluation metrics in order to foster work on automatic methods for wordplay interpretation, generation, and translation.

Abdelrahman Abdallah has joined DS group as a PhD student starting his work from October 1st, 2022. Abdelrahman will conduct research on automatic complex question answering. He has done in the past the research on handwritten text recognition as well as automatic question answering from medical documents.

We are pleased to release an online service for estimating the readability of text in German: Ablesbarkeitsmesser. The website has been developed as a Bachelor's degree project at IFI by: Florian Pickelmann. It is described in the paper which will appear at ECIR 2023 conference [1].

[1] Florian Pickelmann, Michael Färber and Adam Jatowt: Ablesbarkeitsmesser: A System for Assessing the Readability of German Text. Proceedings of the 45th European Conference on Information Retrieval (ECIR 2023), Springer LNCS (to appear in 2023)

The 29th International Conference on Neural Information Processing (**ICONIP 2022**) conference where Adam Jatowt serves as the Program Committee co-chair has attracted over 1,000 submissions.

Bhawna Piryani has joined DS group as a PhD student starting her work from September 1st, 2022. Bhawna will conduct research on novel question answering methods. Bhawna has done in the past research on sentiment analysis.

The survey paper by a master student at IFI, Marco Salchner, on “Automatic Text Summarization using Graph Neural Networks” [1] has been accepted at the 29th International Conference of Computational Linguistics (COLING 2022) to be held in Gyeongju, Republic of South Korea in October 2022.

[1] Marco Salchner, Adam Jatowt: A Survey of Automatic Text Summarization using Graph Neural Networks, Proceedings of the 29th International Conference on Computational Linguistics (COLING 2022), ACL Proceedings (to appear in 2022)

Michael Färber who is the deputy professor (W3) of the research group “Web Science” at KIT will stay with us from 1st August, 2022 to September 30, 2022 as a guest professor. Michael has been selected for UIBK's Guest Professorship program. He works on topics related to the application of semantic web technologies, knowledge graphs and knowledge extraction from scholarly publications.

Adam Jatowt has won the prestigious International Excellence Fellowship from the Karlsruhe Institute of Technology (KIT) which allows for 2-4 months' long research visit at KIT and which will help to strengthen the collaboration with KIT scholars.

Nicolas Gutehrle who is a doctoral student at the University of Franche-Comté stayed with us from 4th April, 2022 for three months as a visiting research student. Nicolas was working on topics related to digital humanities. During his stay at DS group he investigated the application of automatic timeline summarization methods for primary resources.

A paper on large-scale question answering dataset has been accepted to appear at the resource track of the SIGIR 2022 conference [1].

[1] Jiexin Wang, Adam Jatowt, Masatoshi Yoshikawa: ArchivalQA: A Large-scale Benchmark Dataset for Open Domain Question Answering over Historical News Collections. Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2022), ACM Press, (to appear in 2022), preprint available PDF

Adam Jatowt gave a keynote talk entitled: “Temporal Question Answering in News Article Collections” and took part in a panel discussion at the 12th Temporal Web Analytics Workshop (TempWeb 2022), which was held in conjunction with The Web Conference 2022 on April 25/26, Lyon, France.

A paper titled “: Fact-Tree Reasoning for N-ary Question Answering over Knowledge Graphs” [1] has been accepted as a findings paper at the 60th Annual Meeting of the Association for Computational Linguistics (ACL 2022) to be held in Dublin, Ireland in May.

[1] Yao Zhang, Peiyao Li, Hongru Liang, Adam Jatowt, Zhenglu Yang: Fact-Tree Reasoning for N-ary Question Answering over Knowledge Graphs. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL 2022), to appear as ACL findings paper

Our paper titled “Estimating Contemporary Relevance of Past News” [1] has received the Vannevar Bush Best Paper Award which was awarded at the 21tst ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL 2021). This award is co-sponsored by ACM SIGIR group (the ACM Special Interest Group on Information Retrieval) together with SIGWEB (the ACM Special Interest Group on Hypertext and the Web).

[1] Mari Sato, Adam Jatowt, Yijun Duan, Ricardo Campos, Masatoshi Yoshikawa: Estimating Contemporary Relevance of Past News, Proceedings of the 21tst ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL 2021), ACM Press, Illinois, USA, pp. 70-79 (2021)

Following the interview of Adam Jatowt with Fabian Oswald the corresponding news report has appeared at the university newsroom (in German).

A recent research work [1] in collaboration with researchers in Alibaba Group on generating multi-role dialogues has been accepted at the 30th Conference on Information and Knowledge Management (CIKM 2021). The proposed approach can have application as a support for judges in leading case debates and for enhancing customer interactions by chatbot systems.

[1] Changzhen Ji, Yating Zhang, Xiaozhong Liu, Adam Jatowt, Changlong Sun, Conghui Zhu and Tiejun Zhao: A Neural Conversation Generation Model via Equivalent Shared Memory Investigation, Proceedings of the 30th International Conference on Information and Knowledge Management (CIKM 2021), ACM Press (to appear in 2021)

Our survey paper [1] written with researchers from Univ. of La Rochelle on the different approaches used for the correction of OCRed texts has been just published in ACM Computing Surveys journal.

[1] Thi Tuyet Hai Nguyen, Adam Jatowt, Mickaël Coustaty and Antoine Doucet: Survey of Post-OCR Processing Approaches, ACM Computing Surveys, ACM Press (2021) PDF

A new book co-edited with Nina Tahmasebi, Lars Borin, Yang Xu and Simon Hengchen appeared on the topic of computational approaches to semantic change [1]. Semantic change — how the meanings of words change over time — has preoccupied scholars for a long time. Compared to changes in sound and grammar, semantic change is the least understood. Ever since, the study of semantic change has progressed steadily, accumulating a vast store of knowledge for over a century, encompassing many languages and language families. Big-data empirical quantitative investigations are now coming to the forefront, enabled by enormous advances in storage capability and processing power. This volume offers a survey of this exciting new direction in the computational study of semantic change, addressing both more theoretical problems — e.g., discovery of “laws of semantic change” — and practical applications, such as information retrieval in longitudinal text archives.

[1] Nina Tahmasebi, Lars Borin, Adam Jatowt, Yang Xu and Simon Hengchen: Computational Approaches to Semantic Change. Language Variation, Language Science Press (2021)

Multimedia Datasets for Repeatable Experimentation (MDRE 2022) will be held in conjunction with the 27th International Conference on Multimedia Modeling (MMM 2022). MDRE 2022 presents an opportunity for researchers and practitioners to submit their datasets to this track, making their work permanently available and citable via the http://mmdatasets.org forum, as well as to increase the public awareness of their considerable efforts. The special session is organized together with Cathal Gurrin (Dublin City University), Bjorn Por Jonsson (IT University of Copenhagen), Duc-Tien Dang-Nguyen (University of Bergen), Liting Zhou (Dublin City University) and Graham Healy (Dublin City University).

Our paper on multi-timeline summarization has been accepted at the prime NLP conference (ACL-IJCNLP 2021). The paper proposes the new task of generating multiple timeline summaries and demonstrates effective solution for this using two-stage affinity propagation algorithm. This work is the result of collaboration with Kyoto University and University of La Rochelle.

[1] Yu Yi, Adam Jatowt, Antoine Doucet, Kazunari Sugiyama, Masatoshi Yoshikawa: Multi-TimeLine Summarization (MTLS): Improving Timeline Summarization by Generating Multiple Summaries. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing, pp. 377-387 (ACL-IJCNLP 2021) (2021) PDF

The 2nd International Workshop on Computational Approaches to Historical Language Change 2019 ( LChange 2021) was organized together with Nina Tahmasebi (University of Gothenburg), Yang Xu (University of Toronto), Simon Hengchen (University of Gothenburg) Syrielle Montariol (INRIA Paris) and Haim Dubossarsky (University of Cambridge) and was held in conjunction with ACL 2021. The proceedings from the workshop are available at ACL Anthology.

The 6th International Workshop on Computational History ( HistoInformatics 2021) is organized together with Yasunobu Sumikawa (Takushoku University), Ryohei Ikejiri (The University of Tokyo), Antoine Doucet (University of La Rochelle), Eva Pfanzelter (University of Innsbruck), Mohammed Hasanuzzaman (Munster Technological University), Gaël Dias (University of Caen Normandy, GREYC CNRS) and Ian Milligan (University of Waterloo). The workshop will take place in September in conjunction with JCDL 2021 conference.

The 4th workshop on Narrative Extraction from Text (Text2Story 2021) was held in conjunction with ECIR 2021 conference and was organized together with Ricardo Campos (INESC TEC; 2Ci2-Smart Cities Research Center - Polytechnic Institute of Tomar), Alípio M. Jorge (INESC TEC; University of Porto), Sumit Bhatia (IBM Research AI) and Mark Finlayson (Florida International University). The workshop proceedings are available here.

Two papers have been accepted at the main Information Retrieval conference (The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2021)). One paper is about predicting the date of events described in text [1] (collaboration with Kyoto University) using transformers and multi-variate time series analysis and another on multi-modal text summarization (collaboration with IIT Patna).

[1] Jiexin Wang, Adam Jatowt, Masatoshi Yoshikawa: Event Occurrence Date Estimation based on Multivariate Time Series Analysis over Temporal Document Collections. Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2021), ACM Press, pp. 398-407 (2021) PDF

[2] Anubhav Jangra, Sriparna Saha, Adam Jatowt and Mohammed Hasanuzzaman: Multi-Modal Supplementary-Complementary Summarization using Multi-Objective Optimization. Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2021), ACM Press, pp. 818-828 (2021) PDF

The Data Science (DS) group started on January 6, 2021.