Prof. Dr. Andreas Hotho

Head of Data Science Chair and Founding Spokesman of CAIDAS
Data Science Chair (Informatik X)
University of Würzburg
Campus Hubland Nord
Emil-Fischer-Straße 50
97074 Würzburg
Germany

Email: hotho[at]informatik.uni-wuerzburg.de
Phone:(+49 931) 31 - 88453
Mobile: (+49) 173 259 40 52
Office: Room 50.03.004
(Zentrum für Künstliche Intelligenz und Data Science (CAIDAS))
Office Hours: By appointment only

Google Scholar Profile
DBL Profile
Bibsonomy Profile

I am a professor at the University of Würzburg and the head of the Data Science Chair and the founding spokesman of the Center for Artificial Intelligence and Data Science. Prior, I was a senior researcher at the University of Kassel. I started my research at the AIFB Institute at the University of Karlsruhe where I was working on text mining, ontology learning and semantic web related topics. My previous work also involved working at the KDE group of the University of Kassel on topics like data mining, semantic web mining and social media analysis. For a couple of years I've been a member of the L3S Research Center located in Hannover.

I’m a data science expert focusing on developing new data science algorithms and machine learning models for a diverse set of applications and in several interdisciplinary collaborations, which provide interesting challenges for my research. Understanding the models by explainable AI techniques enables my group to effectively build models tailored to the specific challenges of the various application areas.

In the past few years, applying data science and machine learning to ecosystems, environmental & climate data has become one of my central research areas. We have successfully developed deep learning methods for improving climate models in the BigData@Geo project and its successor BigData@Geo 2.0 (jointly with Heiko Paeth) as well as machine learning-based air pollution models in the EveryAware and p2Map project. We’re also analyzing data from smart beehives to understand bee behavior and detect anomalies as swarming events in the we4Bee and BeeConnected (collaboration with Ingolf Steffan-Dewenter) projects.

Another of my major research areas is the work on LLM for Text Mining and NLP in combination with explicitly represented knowledge aka knowledge graphs. Here my group focuses on adapting LLMs and extracting or enriching them with knowledge for our applications, for example in LitBERT to learn more about characters and character networks in novels. We have already worked on methods for representation learning, information extraction, metric and ontology learning and KG enrichment for the Semantic Web and a combination of semantic representations with language models. Specifically, we are developing models for sentiment analysis, scene segmentation and relation detection. With these models, we are able to analyze the development of texts over longer periods: For example, we can follow the plot in fictional novels by tracking the detected relations between characters over scenes, or measure the development of engagement in streams on twitch.tv using sentiment analysis.

To achieve our research objectives, we’re utilizing a rich set of methodological approaches like Knowledge enriched ML, Large Language Models, Time Series and Sequence Modeling, Representation and Metric Learning and Deep Learning for Imbalanced Data, which are described in detail on my group’s research page. For a lot of our research results, we have developed and maintain tools and websites. The most known tools are Bibsonomy, a social bookmark system for publications and We4Bee , a smart beehive monitoring system.

In terms of scientific self-governance, I actively contribute as a PC member, reviewer, and editor across various journals, conferences, and workshops, most recently as an editor in chief for the new diamond open access journal Transactions on Graph Data and Knowledge (TGDK).

Projects

Natural Language Processing und Digital Humanities

LitBERT (DFG, 2023-2026)
KILiMod (BMBF, 2023-2024)
Kallimachos (BMBF, 2014-2017, extended to 2019)
CLiGS (BMBF, 2015 -2019, extended to 2020)
MOTIV (bidt, 2021 - 2023)

ML for Ecosystem and Climate Modeling

BigData@Geo2 (EFRE, 2023-2027)
BeeConnected (2021-2024)
BigData@Geo (EFRE, 2017-2021)
p2map: Learning Environmental Maps (DFG, 2016-2019)
we4Bee (Audi Stiftung 2019 - 2021)
EveryAware: Enhance environmental awareness through social information technologies (EU FET, 2011-2014)

Medical and Biological Data

TissueNet (2023 -2026)
DZ.PTM (2018-2023)

ML for Recommender Systems

DZ.PTM (2018-2023)
REGIO (2018-2021)
adidas (2017-2013)

Security and Fraud

DeepScan (BMBF, 2018 -2021)
Promotionsförderung im Rahmen des Doktorandenprogramms des ZD.B (ZD.B Fellowships, 2017-2020)

Physics Informed Deep Learning

MAGNET (2022-2025)
P-BIM (2022-2024)
AI@Knauf (2019-2023)
KI@FlowChief (2023-2028)

ML for Publication Data

HydrAS (DFG, 2022-2025)
REGIO (BMBF, 2018-2021)
BibSonomy
Pragmatics and Semantics in Social Tagging Systems (DFG, 2011-2016)
PUMA: Academic Publication Management (DFG, 2009-2015)

Industry

AI@Knauf (2019-2023)
Modelling and Recommendation for Customer Engagement (adidas, 2017-2022)
KI@FlowChief (FlowChief, 2023-2028)

Best Paper Honorable Mention: "Developing a Hierarchical Multi-Label Classification Head", Julia Wunderle, Julian Schubert, Antonella Cacciatore, Albin Zehe, Jan Pfister, Andreas Hotho at NAACL 2024
Best Paper Award: "Enhancing Sequential Next-Item Prediction through Modelling Non-Item Pages" , Elisabeth Fischer, Daniel Schlör, Albin Zehe, Andreas Hotho on the Fourth International Workshop on Advanced Neural Algorithms and Theories for Recommender Systems (NeuRec) at ICDM 2023
Best ML Innovation Award: "Deep Learning for Climate Model Output Statistics", Michael Steininger, Daniel Abel, Katrin Ziegler, Anna Krause, Heiko Paeth, Andreas Hotho at Tackling Climate Change with Machine Learning Workshop at NeurIPS 2020 (link)
Best Student Paper Award: "Evaluating the multi-task learning approach for land use regression modelling of air pollution", Andrzej Dulny, Michael Steininger, Florian Lautenschlager, Anna Krause, Andreas Hotho at FAIML 2020
Best Paper Award: "Financial Fraud Detection with Improved Neural Arithmetic Logic Units" by Daniel Schlör, Markus Ring, Anna Krause, Andreas Hotho on the Fifth Workshop on MIning DAta for financial applicationS Co-Hosted by ECML- PKDD 2020
SWSA Ten-Year Award: "Semantic Grounding of Tag Relatedness in Social Bookmarking Systems", Ciro Cattuto, Dominik Benz, Andreas Hotho, Gerd Stumme at the International Semantic Web Conference 2018 (link )
Best Paper Award: "HypTrails: A Bayesian Approach for Comparing Hypotheses About Human Trails on the Web” Philipp Singer, Denis Helic, Andreas Hotho and Markus Strohmaier, at WWW Conference 2015 (link)
Honorable mention of the paper: “Semantic Grounding of Tag Relatedness in Social Bookmarking Systems” Ciro Cattuto, Dominik Benz, Andreas Hotho and Gerd Stumme at ISWC 2008 (link)
The 7 years most influential paper award: “Information Retrieval in Folksonomies: Search and Ranking”, Andreas Hotho, Robert Jäschke, Christoph Schmitz, Gerd Stumme at ESWC 2013 (link )

Current activities:

Executive director: Institute of Computer Science at the University of Würzburg (2013 - 2016, 2023 - 2026)
Founding spokesman of the Center for Artificial Intelligence and Data Science (CAIDAS)
Member of the collegial leadership of the Zentrum für Philologie und Digitalität (Kallimachos)
Spokesperson of the Fachgruppe Knowledge Discovery, Data Mining und Maschinelles Lernen at GI
BAFög coordinator for the computer science department at the University of Würzburg

Editor-in-Chief: Transactions on Graph Data and Knowledge (TGDK) since 2023
Editor: Data Mining and Knowledge Discovery (Journal) since 2023
Senior PC Chair/Area Chair for the Research Track ECML-PKDD
Selected PC memberships: ACM SIGKDD (regularly), AAAI (regularly), WWW (regularly), ISWC (regularly), ESWC (regularly), ECML PKDD (regularly)
Reviewer for journals, e.g., International Journal of Information Security (IJISS), Journal on Data Semantics (JoDS), ACM Transactions on the Web (TWEB), Machine Learning Journal, Data and Knowledge Engineering (DKE)
Reviewer for a variety of workshops
Reviewer of research grants for the DFG and the European Union

Past activities:

Research Track Chair: International Semantic Web Conference 2021
PC Chair:
- ECML - PKDD 2019
- Hypertext 2013
Editor in Chief: Journal of Web Semantics 2018 - 2022
Editorial boards:
- Semantic Web Journal 2009 - 2018
- Journal of Web Semantics (data mining area chair) 2013 -2018
- Transaction on Internet Technology 2013 -2018
Track Co-Chair
- ESWC 2013
- Hypertext 2009, 2011
Demo Co-Chair ECML PKDD 2013
Workshops and Tutorial Chair KCap 2009
Local Co-Chair GI–Workshopwoche “Lernen – Lehren – Wissen – Adaptivität” 2003, 2010
PC Co-Chair for a variety of workshops, e.g.:
- RSWeb at RecSys 2012-2015
- MUSE at ECML PKDD 2010-2015
- Workshop series on semantic web mining at the ECML PKDD 2001- 2005

ModernGBERT: German-only 1B Encoder Model Trained from Scratch A. Ehrmanntraut; J. Wunderle; J. Pfister; F. Jannidis; A. Hotho (2025).
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
@misc{ehrmanntraut2025moderngbert, author = {Ehrmanntraut, Anton and Wunderle, Julia and Pfister, Jan and Jannidis, Fotis and Hotho, Andreas}, keywords = {selected}, title = {ModernGBERT: German-only 1B Encoder Model Trained from Scratch}, year = 2025 }

%0 Generic %1 ehrmanntraut2025moderngbert %A Ehrmanntraut, Anton %A Wunderle, Julia %A Pfister, Jan %A Jannidis, Fotis %A Hotho, Andreas %D 2025 %T ModernGBERT: German-only 1B Encoder Model Trained from Scratch %U https://arxiv.org/abs/2505.13136
LL{\"{a}}Mmlein: Compact and Competitive German-Only Language Models from Scratch J. Pfister; J. Wunderle; A. Hotho in CoRR (2024). abs/2411.11171
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
@article{pfister2024llammlein, author = {Pfister, Jan and Wunderle, Julia and Hotho, Andreas}, journal = {CoRR}, keywords = {selected}, title = {LL{\"{a}}Mmlein: Compact and Competitive German-Only Language Models from Scratch}, volume = {abs/2411.11171}, year = 2024 }

%0 Journal Article %1 pfister2024llammlein %A Pfister, Jan %A Wunderle, Julia %A Hotho, Andreas %D 2024 %J CoRR %R 10.48550/ARXIV.2411.11171 %T LL{\"{a}}Mmlein: Compact and Competitive German-Only Language Models from Scratch %U https://doi.org/10.48550/arXiv.2411.11171 %V abs/2411.11171
ConvMOS: climate model output statistics with deep learning M. Steininger; D. Abel; K. Ziegler; A. Krause; H. Paeth; A. Hotho in Data Mining and Knowledge Discovery (2023). 37(1) 136–166.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
Climate models are the tool of choice for scientists researching climate change. Like all models they suffer from errors, particularly systematic and location-specific representation errors. One way to reduce these errors is model output statistics (MOS) where the model output is fitted to observational data with machine learning. In this work, we assess the use of convolutional Deep Learning climate MOS approaches and present the ConvMOS architecture which is specifically designed based on the observation that there are systematic and location-specific errors in the precipitation estimates of climate models. We apply ConvMOS models to the simulated precipitation of the regional climate model REMO, showing that a combination of per-location model parameters for reducing location-specific errors and global model parameters for reducing systematic errors is indeed beneficial for MOS performance. We find that ConvMOS models can reduce errors considerably and perform significantly better than three commonly used MOS approaches and plain ResNet and U-Net models in most cases. Our results show that non-linear MOS models underestimate the number of extreme precipitation events, which we alleviate by training models specialized towards extreme precipitation events with the imbalanced regression method DenseLoss. While we consider climate MOS, we argue that aspects of ConvMOS may also be beneficial in other domains with geospatial data, such as air pollution modeling or weather forecasts.

@article{steininger2023convmos, abstract = {Climate models are the tool of choice for scientists researching climate change. Like all models they suffer from errors, particularly systematic and location-specific representation errors. One way to reduce these errors is model output statistics (MOS) where the model output is fitted to observational data with machine learning. In this work, we assess the use of convolutional Deep Learning climate MOS approaches and present the ConvMOS architecture which is specifically designed based on the observation that there are systematic and location-specific errors in the precipitation estimates of climate models. We apply ConvMOS models to the simulated precipitation of the regional climate model REMO, showing that a combination of per-location model parameters for reducing location-specific errors and global model parameters for reducing systematic errors is indeed beneficial for MOS performance. We find that ConvMOS models can reduce errors considerably and perform significantly better than three commonly used MOS approaches and plain ResNet and U-Net models in most cases. Our results show that non-linear MOS models underestimate the number of extreme precipitation events, which we alleviate by training models specialized towards extreme precipitation events with the imbalanced regression method DenseLoss. While we consider climate MOS, we argue that aspects of ConvMOS may also be beneficial in other domains with geospatial data, such as air pollution modeling or weather forecasts.}, author = {Steininger, Michael and Abel, Daniel and Ziegler, Katrin and Krause, Anna and Paeth, Heiko and Hotho, Andreas}, journal = {Data Mining and Knowledge Discovery}, keywords = {selected}, number = 1, pages = {136–166}, title = {ConvMOS: climate model output statistics with deep learning}, volume = 37, year = 2023 }

%0 Journal Article %1 steininger2023convmos %A Steininger, Michael %A Abel, Daniel %A Ziegler, Katrin %A Krause, Anna %A Paeth, Heiko %A Hotho, Andreas %D 2023 %J Data Mining and Knowledge Discovery %N 1 %P 136--166 %R 10.1007/s10618-022-00877-6 %T ConvMOS: climate model output statistics with deep learning %U https://doi.org/10.1007/s10618-022-00877-6 %V 37 %X Climate models are the tool of choice for scientists researching climate change. Like all models they suffer from errors, particularly systematic and location-specific representation errors. One way to reduce these errors is model output statistics (MOS) where the model output is fitted to observational data with machine learning. In this work, we assess the use of convolutional Deep Learning climate MOS approaches and present the ConvMOS architecture which is specifically designed based on the observation that there are systematic and location-specific errors in the precipitation estimates of climate models. We apply ConvMOS models to the simulated precipitation of the regional climate model REMO, showing that a combination of per-location model parameters for reducing location-specific errors and global model parameters for reducing systematic errors is indeed beneficial for MOS performance. We find that ConvMOS models can reduce errors considerably and perform significantly better than three commonly used MOS approaches and plain ResNet and U-Net models in most cases. Our results show that non-linear MOS models underestimate the number of extreme precipitation events, which we alleviate by training models specialized towards extreme precipitation events with the imbalanced regression method DenseLoss. While we consider climate MOS, we argue that aspects of ConvMOS may also be beneficial in other domains with geospatial data, such as air pollution modeling or weather forecasts.
InDiReCT: Language-Guided Zero-Shot Deep Metric Learning for Images K. Kobs; M. Steininger; A. Hotho (2022).
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
Common Deep Metric Learning (DML) datasets specify only one notion of similarity, e.g., two images in the Cars196 dataset are deemed similar if they show the same car model. We argue that depending on the application, users of image retrieval systems have different and changing similarity notions that should be incorporated as easily as possible. Therefore, we present Language-Guided Zero-Shot Deep Metric Learning (LanZ-DML) as a new DML setting in which users control the properties that should be important for image representations without training data by only using natural language. To this end, we propose InDiReCT (Image representations using Dimensionality Reduction on CLIP embedded Texts), a model for LanZ-DML on images that exclusively uses a few text prompts for training. InDiReCT utilizes CLIP as a fixed feature extractor for images and texts and transfers the variation in text prompt embeddings to the image embedding space. Extensive experiments on five datasets and overall thirteen similarity notions show that, despite not seeing any images during training, InDiReCT performs better than strong baselines and approaches the performance of fully-supervised models. An analysis reveals that InDiReCT learns to focus on regions of the image that correlate with the desired similarity notion, which makes it a fast to train and easy to use method to create custom embedding spaces only using natural language.

@misc{kobs2022indirect, abstract = {Common Deep Metric Learning (DML) datasets specify only one notion of similarity, e.g., two images in the Cars196 dataset are deemed similar if they show the same car model. We argue that depending on the application, users of image retrieval systems have different and changing similarity notions that should be incorporated as easily as possible. Therefore, we present Language-Guided Zero-Shot Deep Metric Learning (LanZ-DML) as a new DML setting in which users control the properties that should be important for image representations without training data by only using natural language. To this end, we propose InDiReCT (Image representations using Dimensionality Reduction on CLIP embedded Texts), a model for LanZ-DML on images that exclusively uses a few text prompts for training. InDiReCT utilizes CLIP as a fixed feature extractor for images and texts and transfers the variation in text prompt embeddings to the image embedding space. Extensive experiments on five datasets and overall thirteen similarity notions show that, despite not seeing any images during training, InDiReCT performs better than strong baselines and approaches the performance of fully-supervised models. An analysis reveals that InDiReCT learns to focus on regions of the image that correlate with the desired similarity notion, which makes it a fast to train and easy to use method to create custom embedding spaces only using natural language.}, author = {Kobs, Konstantin and Steininger, Michael and Hotho, Andreas}, keywords = {selected}, note = {cite arxiv:2211.12760Comment: Accepted to WACV 2023}, title = {InDiReCT: Language-Guided Zero-Shot Deep Metric Learning for Images}, year = 2022 }

%0 Generic %1 kobs2022indirect %A Kobs, Konstantin %A Steininger, Michael %A Hotho, Andreas %D 2022 %R 10.48550/arXiv.2211.12760 %T InDiReCT: Language-Guided Zero-Shot Deep Metric Learning for Images %U http://arxiv.org/abs/2211.12760 %X Common Deep Metric Learning (DML) datasets specify only one notion of similarity, e.g., two images in the Cars196 dataset are deemed similar if they show the same car model. We argue that depending on the application, users of image retrieval systems have different and changing similarity notions that should be incorporated as easily as possible. Therefore, we present Language-Guided Zero-Shot Deep Metric Learning (LanZ-DML) as a new DML setting in which users control the properties that should be important for image representations without training data by only using natural language. To this end, we propose InDiReCT (Image representations using Dimensionality Reduction on CLIP embedded Texts), a model for LanZ-DML on images that exclusively uses a few text prompts for training. InDiReCT utilizes CLIP as a fixed feature extractor for images and texts and transfers the variation in text prompt embeddings to the image embedding space. Extensive experiments on five datasets and overall thirteen similarity notions show that, despite not seeing any images during training, InDiReCT performs better than strong baselines and approaches the performance of fully-supervised models. An analysis reveals that InDiReCT learns to focus on regions of the image that correlate with the desired similarity notion, which makes it a fast to train and easy to use method to create custom embedding spaces only using natural language.
Detecting Scenes in Fiction: A new Segmentation Task A. Zehe; L. Konle; L. Dümpelmann; E. Gius; A. Hotho; F. Jannidis; L. Kaufmann; M. Krug; F. Puppe; N. Reiter; A. Schreiber; N. Wiedmer (2021).
- [ BibTeX ]
- [ EndNote ]
- [ Download ]
@inproceedings{zehe2021detecting, author = {Zehe, Albin and Konle, Leonard and Dümpelmann, Lea and Gius, Evelyn and Hotho, Andreas and Jannidis, Fotis and Kaufmann, Lucas and Krug, Markus and Puppe, Frank and Reiter, Nils and Schreiber, Annekea and Wiedmer, Nathalie}, booktitle = {Proceedings of the 16th Conference of the {E}uropean Chapter of the Association for Computational Linguistics: Volume 1, Long Papers}, keywords = {selected}, publisher = {ACL}, title = {Detecting Scenes in Fiction: A new Segmentation Task}, year = 2021 }

%0 Conference Paper %1 zehe2021detecting %A Zehe, Albin %A Konle, Leonard %A Dümpelmann, Lea %A Gius, Evelyn %A Hotho, Andreas %A Jannidis, Fotis %A Kaufmann, Lucas %A Krug, Markus %A Puppe, Frank %A Reiter, Nils %A Schreiber, Annekea %A Wiedmer, Nathalie %B Proceedings of the 16th Conference of the {E}uropean Chapter of the Association for Computational Linguistics: Volume 1, Long Papers %D 2021 %I ACL %T Detecting Scenes in Fiction: A new Segmentation Task
Density-based weighting for imbalanced regression M. Steininger; K. Kobs; P. Davidson; A. Krause; A. Hotho in Machine Learning, (A. Appice; S. Escalera; J. A. Gamez; H. Trautmann, Eds.) (2021).
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
In many real world settings,imbalanced data impedes model performance of learning algorithms, like neural networks, mostly for rare cases. This is especially problematic for tasks focusing on these rare occurrences. For example, when estimating precipitation, extreme rainfall events are scarce but important considering their potential consequences. While there are numerous well studied solutions for classification settings, most of them cannot be applied to regression easily. Of the few solutions for regression tasks, barely any have explored cost-sensitive learning which is known to have advantages compared to sampling-based methods in classification tasks. In this work, we propose a sample weighting approach for imbalanced regression datasets called DenseWeight and a cost-sensitive learning approach for neural network regression with imbalanced data called DenseLoss based on our weighting scheme. DenseWeight weights data points according to their target value rarities through kernel density estimation (KDE). DenseLoss adjusts each data point’s influence on the loss according to DenseWeight, giving rare data points more influence on modeltraining compared to common data points. We show on multiple differently distributed datasets that DenseLoss significantly improves model performance for rare data points through its density-based weighting scheme. Additionally, we compare DenseLoss to the state-of-the-art method SMOGN, finding that our method mostly yields better performance. Our approach provides more control over model training as it enables us to actively decide on the trade-off between focusing on common or rare cases through a single hyperparameter, allowing the training of better models for rare data points.

@article{steininger2021densitybased, abstract = {In many real world settings,imbalanced data impedes model performance of learning algorithms, like neural networks, mostly for rare cases. This is especially problematic for tasks focusing on these rare occurrences. For example, when estimating precipitation, extreme rainfall events are scarce but important considering their potential consequences. While there are numerous well studied solutions for classification settings, most of them cannot be applied to regression easily. Of the few solutions for regression tasks, barely any have explored cost-sensitive learning which is known to have advantages compared to sampling-based methods in classification tasks. In this work, we propose a sample weighting approach for imbalanced regression datasets called DenseWeight and a cost-sensitive learning approach for neural network regression with imbalanced data called DenseLoss based on our weighting scheme. DenseWeight weights data points according to their target value rarities through kernel density estimation (KDE). DenseLoss adjusts each data point’s influence on the loss according to DenseWeight, giving rare data points more influence on modeltraining compared to common data points. We show on multiple differently distributed datasets that DenseLoss significantly improves model performance for rare data points through its density-based weighting scheme. Additionally, we compare DenseLoss to the state-of-the-art method SMOGN, finding that our method mostly yields better performance. Our approach provides more control over model training as it enables us to actively decide on the trade-off between focusing on common or rare cases through a single hyperparameter, allowing the training of better models for rare data points.}, author = {Steininger, Michael and Kobs, Konstantin and Davidson, Padraig and Krause, Anna and Hotho, Andreas}, editor = {Appice, Annalisa and Escalera, Sergio and Gamez, Jose A. and Trautmann, Heike}, journal = {Machine Learning}, keywords = {selected}, title = {Density-based weighting for imbalanced regression}, year = 2021 }

%0 Journal Article %1 steininger2021densitybased %A Steininger, Michael %A Kobs, Konstantin %A Davidson, Padraig %A Krause, Anna %A Hotho, Andreas %D 2021 %E Appice, Annalisa %E Escalera, Sergio %E Gamez, Jose A. %E Trautmann, Heike %J Machine Learning %R 10.1007/s10994-021-06023-5 %T Density-based weighting for imbalanced regression %U https://doi.org/10.1007/s10994-021-06023-5 %X In many real world settings,imbalanced data impedes model performance of learning algorithms, like neural networks, mostly for rare cases. This is especially problematic for tasks focusing on these rare occurrences. For example, when estimating precipitation, extreme rainfall events are scarce but important considering their potential consequences. While there are numerous well studied solutions for classification settings, most of them cannot be applied to regression easily. Of the few solutions for regression tasks, barely any have explored cost-sensitive learning which is known to have advantages compared to sampling-based methods in classification tasks. In this work, we propose a sample weighting approach for imbalanced regression datasets called DenseWeight and a cost-sensitive learning approach for neural network regression with imbalanced data called DenseLoss based on our weighting scheme. DenseWeight weights data points according to their target value rarities through kernel density estimation (KDE). DenseLoss adjusts each data point’s influence on the loss according to DenseWeight, giving rare data points more influence on modeltraining compared to common data points. We show on multiple differently distributed datasets that DenseLoss significantly improves model performance for rare data points through its density-based weighting scheme. Additionally, we compare DenseLoss to the state-of-the-art method SMOGN, finding that our method mostly yields better performance. Our approach provides more control over model training as it enables us to actively decide on the trade-off between focusing on common or rare cases through a single hyperparameter, allowing the training of better models for rare data points.
Emote-Controlled: Obtaining Implicit Viewer Feedback through Emote based Sentiment Analysis on Comments of Popular Twitch.tv Channels K. Kobs; A. Zehe; A. Bernstetter; J. Chibane; J. Pfister; J. Tritscher; A. Hotho in {ACM} Transactions on Social Computing (2020). 3(2) 1–34.
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
@article{kobs2020emotecontrolled, author = {Kobs, Konstantin and Zehe, Albin and Bernstetter, Armin and Chibane, Julian and Pfister, Jan and Tritscher, Julian and Hotho, Andreas}, journal = {{ACM} Transactions on Social Computing}, keywords = {selected}, month = {05}, number = 2, pages = {1–34}, publisher = {Association for Computing Machinery ({ACM})}, title = {Emote-Controlled: Obtaining Implicit Viewer Feedback through Emote based Sentiment Analysis on Comments of Popular Twitch.tv Channels}, volume = 3, year = 2020 }

%0 Journal Article %1 kobs2020emotecontrolled %A Kobs, Konstantin %A Zehe, Albin %A Bernstetter, Armin %A Chibane, Julian %A Pfister, Jan %A Tritscher, Julian %A Hotho, Andreas %D 2020 %I Association for Computing Machinery ({ACM}) %J {ACM} Transactions on Social Computing %N 2 %P 1--34 %R 10.1145/3365523 %T Emote-Controlled: Obtaining Implicit Viewer Feedback through Emote based Sentiment Analysis on Comments of Popular Twitch.tv Channels %U https://doi.org/10.1145%2F3365523 %V 3
iNALU: Improved Neural Arithmetic Logic Unit D. Schlör; M. Ring; A. Hotho in Frontiers in Artificial Intelligence (2020). 3 71.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
Neural networks have to capture mathematical relationships in order to learn various tasks. They approximate these relations implicitly and therefore often do not generalize well. The recently proposed Neural Arithmetic Logic Unit (NALU) is a novel neural architecture which is able to explicitly represent the mathematical relationships by the units of the network to learn operations such as summation, subtraction or multiplication. Although NALUs have been shown to perform well on various downstream tasks, an in-depth analysis reveals practical shortcomings by design, such as the inability to multiply or divide negative input values or training stability issues for deeper networks. We address these issues and propose an improved model architecture. We evaluate our model empirically in various settings from learning basic arithmetic operations to more complex functions. Our experiments indicate that our model solves stability issues and outperforms the original NALU model in means of arithmetic precision and convergence.

@article{schlor2020inalu, abstract = {Neural networks have to capture mathematical relationships in order to learn various tasks. They approximate these relations implicitly and therefore often do not generalize well. The recently proposed Neural Arithmetic Logic Unit (NALU) is a novel neural architecture which is able to explicitly represent the mathematical relationships by the units of the network to learn operations such as summation, subtraction or multiplication. Although NALUs have been shown to perform well on various downstream tasks, an in-depth analysis reveals practical shortcomings by design, such as the inability to multiply or divide negative input values or training stability issues for deeper networks. We address these issues and propose an improved model architecture. We evaluate our model empirically in various settings from learning basic arithmetic operations to more complex functions. Our experiments indicate that our model solves stability issues and outperforms the original NALU model in means of arithmetic precision and convergence.}, author = {Schlör, Daniel and Ring, Markus and Hotho, Andreas}, journal = {Frontiers in Artificial Intelligence}, keywords = {selected}, pages = 71, title = {iNALU: Improved Neural Arithmetic Logic Unit}, volume = 3, year = 2020 }

%0 Journal Article %1 schlor2020inalu %A Schlör, Daniel %A Ring, Markus %A Hotho, Andreas %D 2020 %J Frontiers in Artificial Intelligence %P 71 %R 10.3389/frai.2020.00071 %T iNALU: Improved Neural Arithmetic Logic Unit %U https://www.frontiersin.org/article/10.3389/frai.2020.00071 %V 3 %X Neural networks have to capture mathematical relationships in order to learn various tasks. They approximate these relations implicitly and therefore often do not generalize well. The recently proposed Neural Arithmetic Logic Unit (NALU) is a novel neural architecture which is able to explicitly represent the mathematical relationships by the units of the network to learn operations such as summation, subtraction or multiplication. Although NALUs have been shown to perform well on various downstream tasks, an in-depth analysis reveals practical shortcomings by design, such as the inability to multiply or divide negative input values or training stability issues for deeper networks. We address these issues and propose an improved model architecture. We evaluate our model empirically in various settings from learning basic arithmetic operations to more complex functions. Our experiments indicate that our model solves stability issues and outperforms the original NALU model in means of arithmetic precision and convergence.
LM4KG: Improving Common Sense Knowledge Graphs with Language Models J. Omeliyanenko; A. Zehe; L. Hettinger; A. Hotho J. Z. Pan, V. Tamma, C. d’Amato, K. Janowicz, B. Fu, A. Polleres, O. Seneviratne, L. Kagal (Eds.) (2020). 456–473.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
Language Models (LMs) and Knowledge Graphs (KGs) are both active research areas in Machine Learning and Semantic Web. While LMs have brought great improvements for many downstream tasks on their own, they are often combined with KGs providing additionally aggregated, well structured knowledge. Usually, this is done by leveraging KGs to improve LMs. But what happens if we turn this around and use LMs to improve KGs?

@inproceedings{omeliyanenko2020lm4kg, abstract = {Language Models (LMs) and Knowledge Graphs (KGs) are both active research areas in Machine Learning and Semantic Web. While LMs have brought great improvements for many downstream tasks on their own, they are often combined with KGs providing additionally aggregated, well structured knowledge. Usually, this is done by leveraging KGs to improve LMs. But what happens if we turn this around and use LMs to improve KGs?}, address = {Cham}, author = {Omeliyanenko, Janna and Zehe, Albin and Hettinger, Lena and Hotho, Andreas}, booktitle = {The Semantic Web – ISWC 2020}, editor = {Pan, Jeff Z. and Tamma, Valentina and d'Amato, Claudia and Janowicz, Krzysztof and Fu, Bo and Polleres, Axel and Seneviratne, Oshani and Kagal, Lalana}, keywords = {selected}, pages = {456–473}, publisher = {Springer International Publishing}, title = {LM4KG: Improving Common Sense Knowledge Graphs with Language Models}, year = 2020 }

%0 Conference Paper %1 omeliyanenko2020lm4kg %A Omeliyanenko, Janna %A Zehe, Albin %A Hettinger, Lena %A Hotho, Andreas %B The Semantic Web -- ISWC 2020 %C Cham %D 2020 %E Pan, Jeff Z. %E Tamma, Valentina %E d'Amato, Claudia %E Janowicz, Krzysztof %E Fu, Bo %E Polleres, Axel %E Seneviratne, Oshani %E Kagal, Lalana %I Springer International Publishing %P 456--473 %T LM4KG: Improving Common Sense Knowledge Graphs with Language Models %U https://www.informatik.uni-wuerzburg.de/datascience/news/single/news/our-paper-lm4kg-improving-common-sense-knowledge-graphs-with-language-models-has-been-presented-a/ %X Language Models (LMs) and Knowledge Graphs (KGs) are both active research areas in Machine Learning and Semantic Web. While LMs have brought great improvements for many downstream tasks on their own, they are often combined with KGs providing additionally aggregated, well structured knowledge. Usually, this is done by leveraging KGs to improve LMs. But what happens if we turn this around and use LMs to improve KGs? %@ 978-3-030-62419-4
Participatory Patterns in an International Air Quality Monitoring Initiative A. Sîrbu; M. Becker; S. Caminiti; B. De Baets; B. Elen; L. Francis; P. Gravino; A. Hotho; S. Ingarra; V. Loreto; A. Molino; J. Mueller; J. Peters; F. Ricchiuti; F. Saracino; V. D. P. Servedio; G. Stumme; J. Theunis; F. Tria; J. Van den Bossche in PLoS ONE (2015). 10(8) e0136763.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
The issue of sustainability is at the top of the political and societal agenda, being considered of extreme importance and urgency. Human individual action impacts the environment both locally (e.g., local air/water quality, noise disturbance) and globally (e.g., climate change, resource use). Urban environments represent a crucial example, with an increasing realization that the most effective way of producing a change is involving the citizens themselves in monitoring campaigns (a citizen science bottom-up approach). This is possible by developing novel technologies and IT infrastructures enabling large citizen participation. Here, in the wider framework of one of the first such projects, we show results from an international competition where citizens were involved in mobile air pollution monitoring using low cost sensing devices, combined with a web-based game to monitor perceived levels of pollution. Measures of shift in perceptions over the course of the campaign are provided, together with insights into participatory patterns emerging from this study. Interesting effects related to inertia and to direct involvement in measurement activities rather than indirect information exposure are also highlighted, indicating that direct involvement can enhance learning and environmental awareness. In the future, this could result in better adoption of policies towards decreasing pollution.

@article{10.1371/journal.pone.0136763, abstract = {The issue of sustainability is at the top of the political and societal agenda, being considered of extreme importance and urgency. Human individual action impacts the environment both locally (e.g., local air/water quality, noise disturbance) and globally (e.g., climate change, resource use). Urban environments represent a crucial example, with an increasing realization that the most effective way of producing a change is involving the citizens themselves in monitoring campaigns (a citizen science bottom-up approach). This is possible by developing novel technologies and IT infrastructures enabling large citizen participation. Here, in the wider framework of one of the first such projects, we show results from an international competition where citizens were involved in mobile air pollution monitoring using low cost sensing devices, combined with a web-based game to monitor perceived levels of pollution. Measures of shift in perceptions over the course of the campaign are provided, together with insights into participatory patterns emerging from this study. Interesting effects related to inertia and to direct involvement in measurement activities rather than indirect information exposure are also highlighted, indicating that direct involvement can enhance learning and environmental awareness. In the future, this could result in better adoption of policies towards decreasing pollution.}, author = {Sîrbu, Alina and Becker, Martin and Caminiti, Saverio and De Baets, Bernard and Elen, Bart and Francis, Louise and Gravino, Pietro and Hotho, Andreas and Ingarra, Stefano and Loreto, Vittorio and Molino, Andrea and Mueller, Juergen and Peters, Jan and Ricchiuti, Ferdinando and Saracino, Fabio and Servedio, Vito D. P. and Stumme, Gerd and Theunis, Jan and Tria, Francesca and Van den Bossche, Joris}, journal = {PLoS ONE}, keywords = {selected}, month = {08}, number = 8, pages = {e0136763}, publisher = {Public Library of Science}, title = {Participatory Patterns in an International Air Quality Monitoring Initiative}, volume = 10, year = 2015 }

%0 Journal Article %1 10.1371/journal.pone.0136763 %A Sîrbu, Alina %A Becker, Martin %A Caminiti, Saverio %A De Baets, Bernard %A Elen, Bart %A Francis, Louise %A Gravino, Pietro %A Hotho, Andreas %A Ingarra, Stefano %A Loreto, Vittorio %A Molino, Andrea %A Mueller, Juergen %A Peters, Jan %A Ricchiuti, Ferdinando %A Saracino, Fabio %A Servedio, Vito D. P. %A Stumme, Gerd %A Theunis, Jan %A Tria, Francesca %A Van den Bossche, Joris %D 2015 %I Public Library of Science %J PLoS ONE %N 8 %P e0136763 %R 10.1371/journal.pone.0136763 %T Participatory Patterns in an International Air Quality Monitoring Initiative %U http://dx.doi.org/10.1371%2Fjournal.pone.0136763 %V 10 %X The issue of sustainability is at the top of the political and societal agenda, being considered of extreme importance and urgency. Human individual action impacts the environment both locally (e.g., local air/water quality, noise disturbance) and globally (e.g., climate change, resource use). Urban environments represent a crucial example, with an increasing realization that the most effective way of producing a change is involving the citizens themselves in monitoring campaigns (a citizen science bottom-up approach). This is possible by developing novel technologies and IT infrastructures enabling large citizen participation. Here, in the wider framework of one of the first such projects, we show results from an international competition where citizens were involved in mobile air pollution monitoring using low cost sensing devices, combined with a web-based game to monitor perceived levels of pollution. Measures of shift in perceptions over the course of the campaign are provided, together with insights into participatory patterns emerging from this study. Interesting effects related to inertia and to direct involvement in measurement activities rather than indirect information exposure are also highlighted, indicating that direct involvement can enhance learning and environmental awareness. In the future, this could result in better adoption of policies towards decreasing pollution.
Hyptrails: A bayesian approach for comparing hypotheses about human trails P. Singer; D. Helic; A. Hotho; M. Strohmaier (2015).
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
@inproceedings{singer2015hyptrails, address = {Firenze, Italy}, author = {Singer, P. and Helic, D. and Hotho, A. and Strohmaier, M.}, booktitle = {24th International World Wide Web Conference (WWW2015)}, keywords = {selected}, month = {05}, organization = {ACM}, publisher = {ACM}, title = {Hyptrails: A bayesian approach for comparing hypotheses about human trails}, year = 2015 }

%0 Conference Paper %1 singer2015hyptrails %A Singer, P. %A Helic, D. %A Hotho, A. %A Strohmaier, M. %B 24th International World Wide Web Conference (WWW2015) %C Firenze, Italy %D 2015 %I ACM %T Hyptrails: A bayesian approach for comparing hypotheses about human trails %U http://www.www2015.it/documents/proceedings/proceedings/p1003.pdf
Awareness and Learning in Participatory Noise Sensing M. Becker; S. Caminiti; D. Fiorella; L. Francis; P. Gravino; M. (Muki) Haklay; A. Hotho; V. Loreto; J. Mueller; F. Ricchiuti; V. D. P. Servedio; A. Sîrbu; F. Tria in PLoS ONE (2013). 8(12) e81638.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
The development of ICT infrastructures has facilitated the emergence of new paradigms for looking at society and the environment over the last few years. Participatory environmental sensing, i.e. directly involving citizens in environmental monitoring, is one example, which is hoped to encourage learning and enhance awareness of environmental issues. In this paper, an analysis of the behaviour of individuals involved in noise sensing is presented. Citizens have been involved in noise measuring activities through the WideNoise smartphone application. This application has been designed to record both objective (noise samples) and subjective (opinions, feelings) data. The application has been open to be used freely by anyone and has been widely employed worldwide. In addition, several test cases have been organised in European countries. Based on the information submitted by users, an analysis of emerging awareness and learning is performed. The data show that changes in the way the environment is perceived after repeated usage of the application do appear. Specifically, users learn how to recognise different noise levels they are exposed to. Additionally, the subjective data collected indicate an increased user involvement in time and a categorisation effect between pleasant and less pleasant environments.

@article{10.1371/journal.pone.0081638, abstract = {The development of ICT infrastructures has facilitated the emergence of new paradigms for looking at society and the environment over the last few years. Participatory environmental sensing, i.e. directly involving citizens in environmental monitoring, is one example, which is hoped to encourage learning and enhance awareness of environmental issues. In this paper, an analysis of the behaviour of individuals involved in noise sensing is presented. Citizens have been involved in noise measuring activities through the WideNoise smartphone application. This application has been designed to record both objective (noise samples) and subjective (opinions, feelings) data. The application has been open to be used freely by anyone and has been widely employed worldwide. In addition, several test cases have been organised in European countries. Based on the information submitted by users, an analysis of emerging awareness and learning is performed. The data show that changes in the way the environment is perceived after repeated usage of the application do appear. Specifically, users learn how to recognise different noise levels they are exposed to. Additionally, the subjective data collected indicate an increased user involvement in time and a categorisation effect between pleasant and less pleasant environments.}, author = {Becker, Martin and Caminiti, Saverio and Fiorella, Donato and Francis, Louise and Gravino, Pietro and Haklay, Mordechai (Muki) and Hotho, Andreas and Loreto, Vittorio and Mueller, Juergen and Ricchiuti, Ferdinando and Servedio, Vito D. P. and Sîrbu, Alina and Tria, Francesca}, journal = {PLoS ONE}, keywords = {selected}, month = 12, number = 12, pages = {e81638}, publisher = {Public Library of Science}, title = {Awareness and Learning in Participatory Noise Sensing}, volume = 8, year = 2013 }

%0 Journal Article %1 10.1371/journal.pone.0081638 %A Becker, Martin %A Caminiti, Saverio %A Fiorella, Donato %A Francis, Louise %A Gravino, Pietro %A Haklay, Mordechai (Muki) %A Hotho, Andreas %A Loreto, Vittorio %A Mueller, Juergen %A Ricchiuti, Ferdinando %A Servedio, Vito D. P. %A Sîrbu, Alina %A Tria, Francesca %D 2013 %I Public Library of Science %J PLoS ONE %N 12 %P e81638 %R 10.1371/journal.pone.0081638 %T Awareness and Learning in Participatory Noise Sensing %U http://dx.doi.org/10.1371%2Fjournal.pone.0081638 %V 8 %X The development of ICT infrastructures has facilitated the emergence of new paradigms for looking at society and the environment over the last few years. Participatory environmental sensing, i.e. directly involving citizens in environmental monitoring, is one example, which is hoped to encourage learning and enhance awareness of environmental issues. In this paper, an analysis of the behaviour of individuals involved in noise sensing is presented. Citizens have been involved in noise measuring activities through the WideNoise smartphone application. This application has been designed to record both objective (noise samples) and subjective (opinions, feelings) data. The application has been open to be used freely by anyone and has been widely employed worldwide. In addition, several test cases have been organised in European countries. Based on the information submitted by users, an analysis of emerging awareness and learning is performed. The data show that changes in the way the environment is perceived after repeated usage of the application do appear. Specifically, users learn how to recognise different noise levels they are exposed to. Additionally, the subjective data collected indicate an increased user involvement in time and a categorisation effect between pleasant and less pleasant environments.
Collective Information Extraction with Context-Specific Consistencies. P. Klügl; M. Toepfer; F. Lemmerich; A. Hotho; F. Puppe in Lecture Notes in Computer Science, P. A. Flach, T. D. Bie, N. Cristianini (Eds.) (2012). (Vol. 7523) 728–743.
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
@inproceedings{conf/pkdd/KluglTLHP12, author = {Klügl, Peter and Toepfer, Martin and Lemmerich, Florian and Hotho, Andreas and Puppe, Frank}, booktitle = {ECML/PKDD (1)}, crossref = {conf/pkdd/2012-1}, editor = {Flach, Peter A. and Bie, Tijl De and Cristianini, Nello}, keywords = {selected}, pages = {728-743}, publisher = {Springer}, series = {Lecture Notes in Computer Science}, title = {Collective Information Extraction with Context-Specific Consistencies.}, volume = 7523, year = 2012 }

%0 Conference Paper %1 conf/pkdd/KluglTLHP12 %A Klügl, Peter %A Toepfer, Martin %A Lemmerich, Florian %A Hotho, Andreas %A Puppe, Frank %B ECML/PKDD (1) %D 2012 %E Flach, Peter A. %E Bie, Tijl De %E Cristianini, Nello %I Springer %P 728-743 %T Collective Information Extraction with Context-Specific Consistencies. %U http://dblp.uni-trier.de/db/conf/pkdd/pkdd2012-1.html#KluglTLHP12 %V 7523 %@ 978-3-642-33459-7
The Social Bookmark and Publication Management System BibSonomy D. Benz; A. Hotho; R. Jäschke; B. Krause; F. Mitzlaff; C. Schmitz; G. Stumme in The VLDB Journal (2010). 19(6) 849–875.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
Social resource sharing systems are central elements of the Web 2.0 and use the same kind of lightweight knowledge representation, called folksonomy. Their large user communities and ever-growing networks of user-generated content have made them an attractive object of investigation for researchers from different disciplines like Social Network Analysis, Data Mining, Information Retrieval or Knowledge Discovery. In this paper, we summarize and extend our work on different aspects of this branch of Web 2.0 research, demonstrated and evaluated within our own social bookmark and publication sharing system BibSonomy, which is currently among the three most popular systems of its kind. We structure this presentation along the different interaction phases of a user with our system, coupling the relevant research questions of each phase with the corresponding implementation issues. This approach reveals in a systematic fashion important aspects and results of the broad bandwidth of folksonomy research like capturing of emergent semantics, spam detection, ranking algorithms, analogies to search engine log data, personalized tag recommendations and information extraction techniques. We conclude that when integrating a real-life application like BibSonomy into research, certain constraints have to be considered; but in general, the tight interplay between our scientific work and the running system has made BibSonomy a valuable platform for demonstrating and evaluating Web 2.0 research.

@article{benz2010social, abstract = {Social resource sharing systems are central elements of the Web 2.0 and use the same kind of lightweight knowledge representation, called folksonomy. Their large user communities and ever-growing networks of user-generated content have made them an attractive object of investigation for researchers from different disciplines like Social Network Analysis, Data Mining, Information Retrieval or Knowledge Discovery. In this paper, we summarize and extend our work on different aspects of this branch of Web 2.0 research, demonstrated and evaluated within our own social bookmark and publication sharing system BibSonomy, which is currently among the three most popular systems of its kind. We structure this presentation along the different interaction phases of a user with our system, coupling the relevant research questions of each phase with the corresponding implementation issues. This approach reveals in a systematic fashion important aspects and results of the broad bandwidth of folksonomy research like capturing of emergent semantics, spam detection, ranking algorithms, analogies to search engine log data, personalized tag recommendations and information extraction techniques. We conclude that when integrating a real-life application like BibSonomy into research, certain constraints have to be considered; but in general, the tight interplay between our scientific work and the running system has made BibSonomy a valuable platform for demonstrating and evaluating Web 2.0 research.}, address = {Berlin / Heidelberg}, author = {Benz, Dominik and Hotho, Andreas and Jäschke, Robert and Krause, Beate and Mitzlaff, Folke and Schmitz, Christoph and Stumme, Gerd}, journal = {The VLDB Journal}, keywords = {selected}, month = 12, number = 6, pages = {849–875}, publisher = {Springer}, title = {The Social Bookmark and Publication Management System BibSonomy}, volume = 19, year = 2010 }

%0 Journal Article %1 benz2010social %A Benz, Dominik %A Hotho, Andreas %A Jäschke, Robert %A Krause, Beate %A Mitzlaff, Folke %A Schmitz, Christoph %A Stumme, Gerd %C Berlin / Heidelberg %D 2010 %I Springer %J The VLDB Journal %N 6 %P 849--875 %R 10.1007/s00778-010-0208-4 %T The Social Bookmark and Publication Management System BibSonomy %U http://www.kde.cs.uni-kassel.de/pub/pdf/benz2010social.pdf %V 19 %X Social resource sharing systems are central elements of the Web 2.0 and use the same kind of lightweight knowledge representation, called folksonomy. Their large user communities and ever-growing networks of user-generated content have made them an attractive object of investigation for researchers from different disciplines like Social Network Analysis, Data Mining, Information Retrieval or Knowledge Discovery. In this paper, we summarize and extend our work on different aspects of this branch of Web 2.0 research, demonstrated and evaluated within our own social bookmark and publication sharing system BibSonomy, which is currently among the three most popular systems of its kind. We structure this presentation along the different interaction phases of a user with our system, coupling the relevant research questions of each phase with the corresponding implementation issues. This approach reveals in a systematic fashion important aspects and results of the broad bandwidth of folksonomy research like capturing of emergent semantics, spam detection, ranking algorithms, analogies to search engine log data, personalized tag recommendations and information extraction techniques. We conclude that when integrating a real-life application like BibSonomy into research, certain constraints have to be considered; but in general, the tight interplay between our scientific work and the running system has made BibSonomy a valuable platform for demonstrating and evaluating Web 2.0 research.
Tag Recommendations in Social Bookmarking Systems R. Jäschke; L. Marinho; A. Hotho; L. Schmidt-Thieme; G. Stumme in AI Communications, (E. Giunchiglia, Ed.) (2008). 21(4) 231–247.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
Collaborative tagging systems allow users to assign keywords - so called "tags" - to resources. Tags are used for navigation, finding resources and serendipitous browsing and thus provide an immediate benefit for users. These systems usually include tag recommendation mechanisms easing the process of finding good tags for a resource, but also consolidating the tag vocabulary across users. In practice, however, only very basic recommendation strategies are applied. In this paper we evaluate and compare several recommendation algorithms on large-scale real life datasets: an adaptation of user-based collaborative filtering, a graph-based recommender built on top of the FolkRank algorithm, and simple methods based on counting tag occurences. We show that both FolkRank and Collaborative Filtering provide better results than non-personalized baseline methods. Moreover, since methods based on counting tag occurrences are computationally cheap, and thus usually preferable for real time scenarios, we discuss simple approaches for improving the performance of such methods. We show, how a simple recommender based on counting tags from users and resources can perform almost as good as the best recommender.

@article{jaeschke2008tag, abstract = {Collaborative tagging systems allow users to assign keywords - so called "tags" - to resources. Tags are used for navigation, finding resources and serendipitous browsing and thus provide an immediate benefit for users. These systems usually include tag recommendation mechanisms easing the process of finding good tags for a resource, but also consolidating the tag vocabulary across users. In practice, however, only very basic recommendation strategies are applied. In this paper we evaluate and compare several recommendation algorithms on large-scale real life datasets: an adaptation of user-based collaborative filtering, a graph-based recommender built on top of the FolkRank algorithm, and simple methods based on counting tag occurences. We show that both FolkRank and Collaborative Filtering provide better results than non-personalized baseline methods. Moreover, since methods based on counting tag occurrences are computationally cheap, and thus usually preferable for real time scenarios, we discuss simple approaches for improving the performance of such methods. We show, how a simple recommender based on counting tags from users and resources can perform almost as good as the best recommender.}, address = {Amsterdam}, author = {Jäschke, Robert and Marinho, Leandro and Hotho, Andreas and Schmidt-Thieme, Lars and Stumme, Gerd}, editor = {Giunchiglia, Enrico}, journal = {AI Communications}, keywords = {selected}, number = 4, pages = {231-247}, publisher = {IOS Press}, title = {Tag Recommendations in Social Bookmarking Systems}, volume = 21, year = 2008 }

%0 Journal Article %1 jaeschke2008tag %A Jäschke, Robert %A Marinho, Leandro %A Hotho, Andreas %A Schmidt-Thieme, Lars %A Stumme, Gerd %C Amsterdam %D 2008 %E Giunchiglia, Enrico %I IOS Press %J AI Communications %N 4 %P 231-247 %R 10.3233/AIC-2008-0438 %T Tag Recommendations in Social Bookmarking Systems %U http://dx.doi.org/10.3233/AIC-2008-0438 %V 21 %X Collaborative tagging systems allow users to assign keywords - so called "tags" - to resources. Tags are used for navigation, finding resources and serendipitous browsing and thus provide an immediate benefit for users. These systems usually include tag recommendation mechanisms easing the process of finding good tags for a resource, but also consolidating the tag vocabulary across users. In practice, however, only very basic recommendation strategies are applied. In this paper we evaluate and compare several recommendation algorithms on large-scale real life datasets: an adaptation of user-based collaborative filtering, a graph-based recommender built on top of the FolkRank algorithm, and simple methods based on counting tag occurences. We show that both FolkRank and Collaborative Filtering provide better results than non-personalized baseline methods. Moreover, since methods based on counting tag occurrences are computationally cheap, and thus usually preferable for real time scenarios, we discuss simple approaches for improving the performance of such methods. We show, how a simple recommender based on counting tags from users and resources can perform almost as good as the best recommender.
Learning Ontologies to Improve Text Clustering and Classification S. Bloehdorn; P. Cimiano; A. Hotho in From Data and Information Analysis to Knowledge Engineering (2006). 334–341.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
Recent work has shown improvements in text clustering and classification tasks by integrating conceptual features extracted from ontologies. In this paper we present text mining experiments in the medical domain in which the ontological structures used are acquired automatically in an unsupervised learning process from the text corpus in question. We compare results obtained using the automatically learned ontologies with those obtained using manually engineered ones. Our results show that both types of ontologies improve results on text clustering and classification tasks, whereby the automatically acquired ontologies yield a improvement competitive with the manually engineered ones. ER -

@incollection{bloehdorn2006learning, abstract = {Recent work has shown improvements in text clustering and classification tasks by integrating conceptual features extracted from ontologies. In this paper we present text mining experiments in the medical domain in which the ontological structures used are acquired automatically in an unsupervised learning process from the text corpus in question. We compare results obtained using the automatically learned ontologies with those obtained using manually engineered ones. Our results show that both types of ontologies improve results on text clustering and classification tasks, whereby the automatically acquired ontologies yield a improvement competitive with the manually engineered ones. ER -}, author = {Bloehdorn, Stephan and Cimiano, Philipp and Hotho, Andreas}, booktitle = {From Data and Information Analysis to Knowledge Engineering}, keywords = {selected}, pages = {334–341}, publisher = {Springer Berlin Heidelberg}, title = {Learning Ontologies to Improve Text Clustering and Classification}, year = 2006 }

%0 Book Section %1 bloehdorn2006learning %A Bloehdorn, Stephan %A Cimiano, Philipp %A Hotho, Andreas %B From Data and Information Analysis to Knowledge Engineering %D 2006 %I Springer Berlin Heidelberg %P 334--341 %R http://dx.doi.org/10.1007/3-540-31314-1_40 %T Learning Ontologies to Improve Text Clustering and Classification %U http://www.kde.cs.uni-kassel.de/hotho/pub/2006/2006-03-gfkl05-bloehdorn-etal-learning-ontologies.pdf %X Recent work has shown improvements in text clustering and classification tasks by integrating conceptual features extracted from ontologies. In this paper we present text mining experiments in the medical domain in which the ontological structures used are acquired automatically in an unsupervised learning process from the text corpus in question. We compare results obtained using the automatically learned ontologies with those obtained using manually engineered ones. Our results show that both types of ontologies improve results on text clustering and classification tasks, whereby the automatically acquired ontologies yield a improvement competitive with the manually engineered ones. ER - %@ 978-3-540-31313-7
Learning Concept Hierarchies from Text Corpora using Formal Concept Analysis P. Cimiano; A. Hotho; S. Staab in Journal on Artificial Intelligence Research (2005). 24 305–339.
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
- [ Download ]
@article{cimiano05learning, author = {Cimiano, Philipp and Hotho, Andreas and Staab, Steffen}, journal = {Journal on Artificial Intelligence Research}, keywords = {selected}, pages = {305-339}, title = {Learning Concept Hierarchies from Text Corpora using Formal Concept Analysis}, volume = 24, year = 2005 }

%0 Journal Article %1 cimiano05learning %A Cimiano, Philipp %A Hotho, Andreas %A Staab, Steffen %D 2005 %J Journal on Artificial Intelligence Research %P 305-339 %T Learning Concept Hierarchies from Text Corpora using Formal Concept Analysis %U http://dblp.uni-trier.de/db/journals/jair/jair24.html#CimianoHS05 %V 24

List of all publications

Current activities:

Past activities:

Bildnachweise