Andreas Kupfer

Improving literature preselection by searching for images

In this paper we present a picture search engine for life science literature and show how it can be used to improve literature preselection. This preselection is needed as a way to compensate for the vast amounts of literature that are available. While searching for DNA binding sites for example, we wanted to add the results of specific experiments (DNAse I footprint and EMSA) to our database. The preselection via abstract search was very unspecific (150 000 hits), but by looking for paper with images concerning the experiments, we could improve precision immensely. They are displayed like hits in a search engine, allowing easy and quick quality assessment without having to read through the whole paper. The images are found by their annotation in the paper: the figure caption. To identify that, we analyse the layout of the paper: the position of the image and the surrounding text.

Discovering Gene Expression Data from the Tables of Full Text Publications

Finding out which genes are expressed in which circumstances is one of the most common tasks in text mining for bioinformatics. But usually the derived data comes from the abstract or other describing texts in the literature. In the age of modern high-throughput microarray analysis, however, there is too much data to be described textual; instead this data often comes in form of tables. In this paper, we are looking specifically at the tables, an approach to our knowledge never described before. The goal is to attach gene names found in tables to their context for a convenient literature review. In order to do so, matching literature has to be downloaded and pre-processed. After that has been done, gene names or protein names can be found through a fast and reliable search, presenting all the associated literature at a glance.

Handling changes of database schemas and corresponding ontologies

Currently, knowledge from biological research is stored in hundreds of databases, counting only public accessible ones. Finding specific data in these is a challenging task which can be supported by ontologies describing them. The maintenance of a corresponding ontology is time consuming manual work, because research database schemas change rapidly. Our project will reduce the work by automating tasks, like a generation process and applying schema changes to the corresponding ontology. We call the proposed method coevolution, because database schema and ontology are allowed to evolve independently without ever losing their connection to each other. Our method consists of initial ontology generation, manual annotation and change propagation steps.

A database ontology for signal transduction pathways

Ontologies are one of the key technologies for data integration and meta-databases, by connecting databases at a semantical level. Still, the database has to be connected to the ontology and vice versa. To achieve this, we propose a two step process. First, we automatically generate an ontology directly from the database schema. Next, we annotate this with concepts from a domain specific ontology. We demonstrate this for signal transduction pathways, which describe how cells can react to external stimuli. In this paper we present our mapping of database schemas to ontologies and describe how these ontologies can be enriched by semantical information.

Keeping track of changes in database schemas and related ontologies

Connecting scientific databases is a challenging task which can be supported by ontologies describing them on a semantical level. Unfortunately, ontologies for databases are rarely used, because schemas of research databases change rapidly while the ontologies grow as well. Maintaining the connection between both is time consuming manual work. Therefore it is worthwhile to reduce the required work by automating tasks. We propose an approach that allows the database schema and the ontology to change and evolve, without ever losing their connection to each other. This is done by mapping database schemas to ontologies, enriching these ontologies with semantical information and transfering schema changes to the ontology

Modellierung und kartographische Visualisierung von Geodaten mit XML-basierten Sprachen

ZusammenfassungMit den XML-basierten Sprachen GML, XSLT und SVG lassen sich Geodaten nicht nur anwendungsorientiert modellieren, sondern auch kartenähnlich visualisieren. In dieser Fallstudie zeigen wir das, indem wir realistische Geodatenbestände der Landesvermessungsämter zunächst mit der Geography Markup Language (GML) nachmodellieren. So mit GML strukturierte Daten werden dann mit der Extensible Stylesheet Language for Transformation (XSLT) auf Elemente der Sprache Scalable Vector Graphics (SVG) abgebildet. Dabei wird der Prozess der kartografischen Visualisierung durch XSLT-Konstrukte modelliert und auch gleichzeitig implementiert. Als Ergebnis erhalten wir Grafiken, die den entsprechenden Karten der Landesämter zumindest nicht unähnlich sind. AbstractUsing the XML-based languages GML, XSLT and SVG, we modeled German geo data and also visualized it in map-like graphics. This case study shows the feasibility of that approach, by modelling the data with the Geography Markup Language (GML). Then, the GML-structured data is mapped with the Extensible Stylesheet Language for Transformation (XSLT) to elements of the language Scalable Vector Graphics (SVG). The process of cartographical visualization is thus modelled through XSLT-constructs and at same time also implemented. The results are graphics, which share close ressemblance to the corresponding maps of the official offices.

Using Layout Data for the Analysis of Scientific Literature

It is said that the world knowledge is in the Internet. Scientific knowledge is in the books, journals and conference proceedings. Yet both repositories are too large to skim through manually. We need clever algorithms to cope with the huge amount of information. To filter, sort and ultimately mine the information available it is vital to use every source of information we have. A common technique is to mine the text from the publications, but they are more complex than the text they include. The position of the words gives us clues about their meaning. Additional images either supplement the text or offer proof to a proposition. Tables cannot be understood before deciphering the rows and columns. To deal with the additional information, classic text mining techniques have to be coupled with spatial data and image data. In this chapter, we will give some background to the various techniques, explain the necessary pre-processing steps involved and present two case studies, one from image mining and one from table identification.

Using image classification for biomedical literature retrieval

The aim of literature retrieval is to find significant papers on a given topic. In previous publications, we examined the use of choosing these papers based on the pictures they include. To refine this approach we seek to employ picture classification to further narrow down the number of interesting pictures presented. This can be useful, for example, when looking for the results of specific experiments. The classification can also be useful as a data cleansing step, to omit all unnecessary pictures not used as a figure. We use a method originally designed to distinguish between photos and computer-generated pictures on the Web. We show that this method can not only be used to distinguish between raw data and derived representation figures, we can also reliably eliminate nonfigure pictures in the document, like text pages and logos. We tested this approach on two different data sets with different topics and different non-figure problems, both with satisfactory results