README.md 6.59 KB
Newer Older
Alexander Hinneburg's avatar
Alexander Hinneburg committed
1
2
# Flash-Talks

Alexander Hinneburg's avatar
Alexander Hinneburg committed
3
4
Einmal im Semester wollen wir im Rahmen des Institutskolloquiums allen Interessierten die aktuell laufenden Forschungsarbeiten am Institut für Informatik vorstellen.
Das soll in einem besonderen Format passieren: den Flash-Talks.
Alexander Hinneburg's avatar
Alexander Hinneburg committed
5

Alexander Hinneburg's avatar
Alexander Hinneburg committed
6
7
8
9
In Flash-Talks können die Vortragenden in Kürze
zeigen, woran sie zur Zeit arbeiten und forschen.
Die DoktorandInnen, PostDocs sowie die Mitarbeitenden und Professoren sind aufgerufen, sich mit Vorträgen zu beteiligen.
Gleiches gilt auch für Master- und Bachelor-Studierende,
10
die in die Forschung der Arbeitsgruppen (HiWi, Abschlussarbeiten, Forschungsprojekte) einbezogen sind. Im Anschluss soll es Gelegenheit zu Gesprächen geben.
Alexander Hinneburg's avatar
Alexander Hinneburg committed
11

Alexander Hinneburg's avatar
Alexander Hinneburg committed
12
13
- **Nächster Termin: Donnerstag 23.1.2020 16.00-18.00 Uhr**
- **Raum: 3.31**
14
- **Deadline für Vortragsanmeldung: Mittwoch 8.1.2020**  
Alexander Hinneburg's avatar
Alexander Hinneburg committed
15
16
17

<details>
  <summary>
18
    Format für Flash-Talks
Alexander Hinneburg's avatar
Alexander Hinneburg committed
19
  </summary>
Alexander Hinneburg's avatar
Alexander Hinneburg committed
20
  <ul>
21
   <li>etwa 5 Minuten Vortrag + 2 Minuten</li>
Alexander Hinneburg's avatar
Alexander Hinneburg committed
22
23
24
25
26
   <li> Vortragssprache wenn möglich Englisch</li>
  </ul>
</details>
<details>
  <summary>
27
    Anmeldung und Einreichung eines Flash-Talks
Alexander Hinneburg's avatar
Alexander Hinneburg committed
28
29
  </summary>
  <ol>
30
31
   <li>Im <a href="https://gitlab.informatik.uni-halle.de/hinnebur/flashtalks">Flashtalks-Projekt</a> im Gitlab mit LDAP-Account einloggen.</li>
   <li>Datei <b>README.md</b> editieren. Es gibt erstmal 10 Slots.</li>
Alexander Hinneburg's avatar
Alexander Hinneburg committed
32
   <li>Bis 8.1.2020 für die Anmeldung einfach die Felder eines freien Slots ausfüllen und Merge-Request abschicken.</li>
33
   <li>Bis 23.1.2020 12 Uhr, PDF-Datei mit Flash-Talk im git hochladen.</li>
Alexander Hinneburg's avatar
Alexander Hinneburg committed
34
35
36
37
  </ol>
</details>


38
## Flash-Talk Programm, Do. 23.1.2020, 16.00 Uhr
Alexander Hinneburg's avatar
Alexander Hinneburg committed
39
40
41
- **Eröffnung**


Maik Fröbe's avatar
Maik Fröbe committed
42
43
44
1. **Maik Fröbe**, Big Data Analytics  
  **Hidden Oversampling in Learning-to-Rank**  
  The machine learning approaches to solve the problem of ranking in search engines rely on vast amounts of training data. Those training datasets contain duplicated documents that are unexpected in subsequent machine learning algorithms. We investigate the impact of that "hidden oversampling" to the produced rankings of the search engine.
Alexander Bondarenko's avatar
Alexander Bondarenko committed
45
46
47
48
49
50
2. **Alexander Bondarenko**, Big Data Analytics  
  **Comparative Web Search Questions**  
  Users ask comparative questions, i.e., questions asking to compare different items, not only on community question answering platforms like Yahoo! Answers, Quora, or StackExchange, but also submit as queries to search engines. Responses to such questions might be quite different from the simple ''ten blue links'' and could, for example, aggregate pros and cons of the different options as direct answers. We analyze such questions and propose methods to answering them.
3. **Ekaterina Shirshakova**, Big Data Analytics  
  **Same Side Stance Classification**  
  In recent years, the popularity of social media and online discussions has lead to a rise of pro and con argumentation for various topics. Still, since not all contributions in such online discussions clearly indicate a stance or polarity of the contribution, automatically identifying some post's stance (in social media platforms, etc.) could help readers quickly get an overview of a discussion similar to debating portals with pro/con arguments.
Mario Wenzel's avatar
Mario Wenzel committed
51
52
53
4. **Mario Wenzel**,  Datenbanken und Informationssysteme   
  **Microlog - Datalog for microcontrollers**  
  Datalog or variants thereof are prevalent in an uncountable number of software systems. An area that seems ideal for rule-based programming, microcontrollers, is underserved by logic programming approaches. Microlog aims to close that gap by providing a standalone framework for deductive reasoning in interactive systems.
Silvio Weging's avatar
Silvio Weging committed
54
55
56
5. **Silvio Weging**, Bioinformatik  
  **kASA: Taxonomic Analysis of Metagenomic Data on a Notebook**  
  The taxonomic analysis of metagenomic sequencing data has become important in many areas of life sciences. However, currently available software tools for that purpose either consume large amounts of RAM or yield an insufficient quality of the results. To identify and profile metagenomic sequences with high computational efficiency and a small user-definable memory footprint, we developed a k-mer based software named "kASA".
Annett Erkes's avatar
Annett Erkes committed
57
58
59
60
6. **Annett Erkes**, Bioinformatik  
  **Bioinformatic analysis of TAL effectors in the Xanthomonas oryzae - rice interaction**  
  Diseases caused by plant-pathogenic Xanthomonas bacteria are a serious threat for many important crop plants including rice. Efficiently protecting plants from these pathogens requires a deeper understanding of infection strategies. Such infection strategies depend on a special class of effector
proteins, termed transcription activator-like effectors (TALEs). Our approach PrediTALE predicts plant target genes of TALEs.
Vaibhav Kasturia's avatar
Vaibhav Kasturia committed
61
62
7. **Vaibhav Kasturia**, Big Data Analytics  
  **Entity-Based Query Interpretation**  
63
  We address the problem of entity-based query interpretation: given a keyword query, what accepted meanings can the query have with respect to potentially ambiguous contained entities. To tackle this problem, we separate three entity recognition problems: explicit entity recognition, implicit entity recognition and related entity recognition. Based on these problems, we define entity-based query interpretation and introduce a new corpus containing 2800 queries with explicit and implicit entities as well as with query interpretations. Using query segmentation, the possible meanings of a query are output based on the entities contained in the query.
64
65
8. **Maximilian Konzack**, Datenbanken und Informationssysteme / iDiv  
  **Information extraction from ecological publications**  
66
  Biodiversity datasets are increasingly deposited either as supplementary material in an article or in online repositories. As more and more scientific papers are digitally published, we face new challenges to find, access, integrate and reuse publications and their associated datasets. In this talk, I will talk about methodologies in which I employ text classification and geoparsing to tackle some of these problems.
67
9. **Marcus Pöckelmann**, Informatik in den Geisteswissenschaften  
Marcus Poeckelmann's avatar
Marcus Poeckelmann committed
68
69
  **Collation and analysis of Hebrew text witnesses**  
  The Hebrew treatise Keter Shem Ṭov (“Crown of the Good Name”) composed in the 13th century is one of the most important introductory texts into the Kabbalah, testified by about 100 differing manuscripts. In our current project, an interdisciplinary team prepares a scholarly edition of this work with the aid of our tool LERA that has to be adopted for this task. The focus here is on the enormous number of text witnesses, which poses a challenge for both the underlying algorithms and the user interface of LERA.
Alexander Hinneburg's avatar
Alexander Hinneburg committed
70
71
72
10. **Name**, Arbeitsgruppe  
  **Vortragstitel**  
  Inhaltliche Kurzbeschreibung in 2-3 Sätzen, wenn möglich in Englisch.