Alexander Dallmann, Dipl. Inf.

Chair of Data Science (Informatik X)
University of Würzburg
Am Hubland
97074 Würzburg
Germany

Email: dallmann[at]informatik.uni-wuerzburg.de

Phone: (+49 931) 31 - 89097

Fax: +49 931 / 31 - 86732

Office: Room 0.003 (Building M4)

Projects and Research Interests

I received my Diploma in Computer Science from the University of Würzburg in 2012. Afterwards I started working on PABS a Programming Assignment Scoring System for the Chair of Computer Science 2 in Würzburg. PABS is used in several courses at the university. In 2014 I started on my PHD in machine learning at the Data Mining and Information Retrieval Group. In my research I focus on deep learning models, but I'm also interested in software development and advances in programming languages.

Publications

2021[ to top ]

Sequential Item Recommendation in the MOBA Game Dota 2. Dallmann, Alexander; Kohlmann, Johannes; Zoller, Daniel; Hotho, Andreas. In 2021 International Conference on Data Mining Workshops (ICDMW), pp. 10–17. 2021.
- [ Abstract ]
- [ BibTeX ]
- [ URL ]
- [ DOI ]
Multiplayer Online Battle Arena (MOBA) games such as Dota 2 attract hundreds of thousands of players every year. Despite the large player base, it is still important to attract new players to prevent the community of a game from becoming inactive. Entering MOBA games is, however, often demanding, requiring the player to learn numerous skills at once. An important factor of success is buying the correct items which forms a complex task depending on various in-game factors such as already purchased items, the team composition, or available resources. A recommendation system can support players by reducing the mental effort required to choose a suitable item, helping, e.g., newer players or players returning to the game after a longer break, to focus on other aspects of the game. Since Sequential Item Recommendation (SIR) has proven to be effective in various domains (e.g. e-commerce, movie recommendation or playlist continuation), we explore the applicability of well-known SIR models in the context of purchase recommendations in Dota 2. To facilitate this research, we collect, analyze and publish Dota-350k, a new large dataset based on recent Dota 2 matches. We find that SIR models can be employed effectively for item recommendation in Dota 2. Our results show that models that consider the order of purchases are the most effective. In contrast to other domains, we find RNN-based models to outperform the more recent Transformer-based architectures on Dota-350k.

@inproceedings{9679858, abstract = {Multiplayer Online Battle Arena (MOBA) games such as Dota 2 attract hundreds of thousands of players every year. Despite the large player base, it is still important to attract new players to prevent the community of a game from becoming inactive. Entering MOBA games is, however, often demanding, requiring the player to learn numerous skills at once. An important factor of success is buying the correct items which forms a complex task depending on various in-game factors such as already purchased items, the team composition, or available resources. A recommendation system can support players by reducing the mental effort required to choose a suitable item, helping, e.g., newer players or players returning to the game after a longer break, to focus on other aspects of the game. Since Sequential Item Recommendation (SIR) has proven to be effective in various domains (e.g. e-commerce, movie recommendation or playlist continuation), we explore the applicability of well-known SIR models in the context of purchase recommendations in Dota 2. To facilitate this research, we collect, analyze and publish Dota-350k, a new large dataset based on recent Dota 2 matches. We find that SIR models can be employed effectively for item recommendation in Dota 2. Our results show that models that consider the order of purchases are the most effective. In contrast to other domains, we find RNN-based models to outperform the more recent Transformer-based architectures on Dota-350k.}, author = {Dallmann, Alexander and Kohlmann, Johannes and Zoller, Daniel and Hotho, Andreas}, booktitle = {2021 International Conference on Data Mining Workshops (ICDMW)}, keywords = {sir}, month = 12, pages = {10-17}, title = {Sequential Item Recommendation in the MOBA Game Dota 2}, year = 2021 }
A Case Study on Sampling Strategies for Evaluating Neural Sequential Item Recommendation Models. Dallmann, Alexander; Zoller, Daniel; Hotho, Andreas. In Fifteenth {ACM} Conference on Recommender Systems. {ACM}, 2021.
- [ BibTeX ]
- [ DOI ]
- [ Download ]
@inproceedings{dallmann2021study, author = {Dallmann, Alexander and Zoller, Daniel and Hotho, Andreas}, booktitle = {Fifteenth {ACM} Conference on Recommender Systems}, keywords = {research:recommender:selected}, month = {09}, publisher = {{ACM}}, title = {A Case Study on Sampling Strategies for Evaluating Neural Sequential Item Recommendation Models}, year = 2021 }

2020[ to top ]

Integrating Keywords into BERT4Rec for Sequential Recommendation. Fischer, Elisabeth; Zoller, Daniel; Dallmann, Alexander; Hotho, Andreas. In KI 2020: Advances in Artificial Intelligence. 2020.
- [ Abstract ]
- [ BibTeX ]
- [ DOI ]
- [ Download ]
A crucial part of recommender systems is to model the user’s preference based on her previous interactions. Different neural networks (e.g., Recurrent Neural Networks), that predict the next item solely based on the sequence of interactions have been successfully applied to sequential recommendation. Recently, BERT4Rec has been proposed, which adapts the BERT architecture based on the Transformer model and training methods used in the Neural Language Modeling community to this task. However, BERT4Rec still only relies on item identifiers to model the user preference, ignoring other sources of information. Therefore, as a first step to include additional information, we propose KeBERT4Rec, a modification of BERT4Rec, which utilizes keyword descriptions of items. We compare two variants for adding keywords to the model on two datasets, a Movielens dataset and a dataset of an online fashion store. First results show that both versions of our model improves the sequential recommending task compared to BERT4Rec.

@inproceedings{fischer2020integrating, abstract = {A crucial part of recommender systems is to model the user’s preference based on her previous interactions. Different neural networks (e.g., Recurrent Neural Networks), that predict the next item solely based on the sequence of interactions have been successfully applied to sequential recommendation. Recently, BERT4Rec has been proposed, which adapts the BERT architecture based on the Transformer model and training methods used in the Neural Language Modeling community to this task. However, BERT4Rec still only relies on item identifiers to model the user preference, ignoring other sources of information. Therefore, as a first step to include additional information, we propose KeBERT4Rec, a modification of BERT4Rec, which utilizes keyword descriptions of items. We compare two variants for adding keywords to the model on two datasets, a Movielens dataset and a dataset of an online fashion store. First results show that both versions of our model improves the sequential recommending task compared to BERT4Rec.}, author = {Fischer, Elisabeth and Zoller, Daniel and Dallmann, Alexander and Hotho, Andreas}, booktitle = {KI 2020: Advances in Artificial Intelligence}, keywords = {selected}, title = {Integrating Keywords into BERT4Rec for Sequential Recommendation}, year = 2020 }

2018[ to top ]

ClaiRE at SemEval-2018 Task 7 - Extended Version. Hettinger, Lena; Dallmann, Alexander; Zehe, Albin; Niebler, Thomas; Hotho, Andreas. 2018.
- [ Abstract ]
- [ BibTeX ]
- [ URL ]
- [ Download ]
In this paper we describe our post-evaluation results for SemEval-2018 Task 7 on classification of semantic relations in scientific literature for clean (subtask 1.1) and noisy data (subtask 1.2). Due to space limitations we publish an extended version of Hettinger et al. (2018) including further technical details and changes made to the preprocessing step in the post-evaluation phase. Due to these changes Classification of Relations using Embeddings (ClaiRE) achieved an improved F1 score of 75.11% for the first subtask and 81.44% for the second.

@misc{hettinger2018claire, abstract = {In this paper we describe our post-evaluation results for SemEval-2018 Task 7 on classification of semantic relations in scientific literature for clean (subtask 1.1) and noisy data (subtask 1.2). Due to space limitations we publish an extended version of Hettinger et al. (2018) including further technical details and changes made to the preprocessing step in the post-evaluation phase. Due to these changes Classification of Relations using Embeddings (ClaiRE) achieved an improved F1 score of 75.11% for the first subtask and 81.44% for the second.}, author = {Hettinger, Lena and Dallmann, Alexander and Zehe, Albin and Niebler, Thomas and Hotho, Andreas}, keywords = {deep_learning}, note = {cite arxiv:1804.05825Comment: This is the extended version for our work: ClaiRE at SemEval-2018 Task 7}, title = {ClaiRE at SemEval-2018 Task 7 - Extended Version}, year = 2018 }

2016[ to top ]

Extracting Semantics from Random Walks on Wikipedia: Comparing learning and counting methods. Dallmann, Alexander; Niebler, Thomas; Lemmerich, Florian; Hotho, Andreas. In Wiki Workshop@ICWSM, R. West, L. Zia, D. Taraborelli, J. Leskovec (eds.). 2016.
- [ Abstract ]
- [ BibTeX ]
- [ URL ]
- [ Download ]
Semantic relatedness between words has been extracted from a variety of sources. In this ongoing work, we explore and compare several options for determining if semantic relatedness can be extracted from navigation structures in Wikipedia. In that direction, we first investigate the potential of representation learning techniques such as DeepWalk in comparison to previously applied methods based on counting co-occurrences. Since both methods are based on (random) paths in the network, we also study different approaches to generate paths from Wikipedia link structure. For this task, we do not only consider the link structure of Wikipedia, but also actual navigation behavior of users. Finally, we analyze if semantics can also be extracted from smaller subsets of the Wikipedia link network. As a result we find that representa- tion learning techniques mostly outperform the investigated co-occurrence counting methods on the Wikipedia network. However, we find that this is not the case for paths sampled from human navigation behavior.

@inproceedings{dallmann2016extracting, abstract = {Semantic relatedness between words has been extracted from a variety of sources. In this ongoing work, we explore and compare several options for determining if semantic relatedness can be extracted from navigation structures in Wikipedia. In that direction, we first investigate the potential of representation learning techniques such as DeepWalk in comparison to previously applied methods based on counting co-occurrences. Since both methods are based on (random) paths in the network, we also study different approaches to generate paths from Wikipedia link structure. For this task, we do not only consider the link structure of Wikipedia, but also actual navigation behavior of users. Finally, we analyze if semantics can also be extracted from smaller subsets of the Wikipedia link network. As a result we find that representa- tion learning techniques mostly outperform the investigated co-occurrence counting methods on the Wikipedia network. However, we find that this is not the case for paths sampled from human navigation behavior.}, author = {Dallmann, Alexander and Niebler, Thomas and Lemmerich, Florian and Hotho, Andreas}, booktitle = {Wiki Workshop@ICWSM}, editor = {West, Robert and Zia, Leila and Taraborelli, Dario and Leskovec, Jure}, keywords = {wikipedia}, title = {Extracting Semantics from Random Walks on Wikipedia: Comparing learning and counting methods}, year = 2016 }

2015[ to top ]

Media Bias in German Online Newspapers. Dallmann, Alexander; Lemmerich, Florian; Zoller, Daniel; Hotho, Andreas. In 26th ACM Conference on Hypertext and Social Media. ACM, Cyprus, Turkey, September 1-4, 2015.
- [ BibTeX ]
@inproceedings{dallmann2015media, address = {Cyprus, Turkey, September 1-4}, author = {Dallmann, Alexander and Lemmerich, Florian and Zoller, Daniel and Hotho, Andreas}, booktitle = {26th ACM Conference on Hypertext and Social Media}, keywords = {media}, publisher = {ACM}, title = {Media Bias in German Online Newspapers}, year = 2015 }

2014[ to top ]

Finding Enclosures for Linear Systems Using Interval Matrix Multiplication in CUDA. Dallmann, Alexander; Beck, Philip-Daniel; von Gudenberg, JürgenWolff. In Parallel Processing and Applied Mathematics, Vol. 8385, R. Wyrzykowski, J. Dongarra, K. Karczewski, J. Waśniewski (eds.), pp. 582–590. Springer Berlin Heidelberg, 2014.
- [ Abstract ]
- [ BibTeX ]
- [ URL ]
- [ DOI ]
- [ Download ]
In this paper we present CUDA kernels that compute an interval matrix product. Starting from a naive implementation we investigate possible speedups using commonly known techniques from standard matrix multiplication. We also evaluate the achieved speedup when our kernels are used to accelerate a variant of an existing algorithm that finds an enclosure for the solution of a linear system. Moreover the quality of our enclosure is discussed.

@incollection{noKey, abstract = {In this paper we present CUDA kernels that compute an interval matrix product. Starting from a naive implementation we investigate possible speedups using commonly known techniques from standard matrix multiplication. We also evaluate the achieved speedup when our kernels are used to accelerate a variant of an existing algorithm that finds an enclosure for the solution of a linear system. Moreover the quality of our enclosure is discussed.}, author = {Dallmann, Alexander and Beck, Philip-Daniel and von Gudenberg, JürgenWolff}, booktitle = {Parallel Processing and Applied Mathematics}, editor = {Wyrzykowski, Roman and Dongarra, Jack and Karczewski, Konrad and Waśniewski, Jerzy}, keywords = {enclosures}, pages = {582-590}, publisher = {Springer Berlin Heidelberg}, series = {Lecture Notes in Computer Science}, title = {Finding Enclosures for Linear Systems Using Interval Matrix Multiplication in CUDA}, volume = 8385, year = 2014 }

Hubland Nord

Bildnachweise