praise from caesar

From Peter Murray-Rust’s enthusiastic and witty blog:

“I’ve worked with Soton and Indiana before, and it has been great to make new and excting links with PSU. Many of you will know them for CiteSeer and ChemXSeer but I hadn’t realised how well their information extraction techniques mapped onto ours. They’ve got some pretty smart stuff for extracting information from chemistry in PDF documents which maps directly onto OSCAR3+OPSIN and CML. “

DocEng09 paper (almost done)

Abstract: For at least fifty years, liquid state Nuclear Magnetic Resonance (NMR) has served as an important analytical technique in studying local atomic bonding information. Thus, a vast amount of data of interest to the chemist and crystallographer resides in archived documents, containing liquid state NMR spectra and accompanying molecular structures. These structures are determined on the basis of chemical shift information from spectra, using well established empirical rules. The combined wealth of information represented visually in the spectra and molecules precludes straightforward inclusion in a traditional database. Given its value to the researcher, work by this group is being dedicated to automatic extraction of spectral and molecular information from documents, for conversion to the Chemical Markup Language (CML) and incorporation into a database. Preliminary results are presented here, as well as details of future work.

Keywords: Information Extraction, Chemical Markup Language, Nuclear Magnetic Resonance, Computer Assisted Structure Determination, PostScript

Mining Spectra + Molecules

I’ve been working with colleagues and collaborators from the UK to mine NMR spectra and corresponding molecular structures from documents. The object is to create an XML/CML database to give researchers unprecedented access to information, useful in (for instance) drug discovery. At this stage, I have focused on writing algorithms for the extraction of molecules to *svg, and NMR data to *txt. The latter is then refined and processed, first to determine peak positions. The data is then optimally fit using mixture models, and peak lists created automatically using standard and novel algorithms.

Press for Image Proc Work

In an unusual twist, a few news sources are reporting on the aforementioned image proc work at Penn State: New Scientist , Naked Scientists and L’Atelier

I’m flattered and more than a little surprised

ED: also showed up in ACM tech news

JCDL image extraction paper/talk

Here’s the paper abstract, links to paper/talk follow:

“Most search engines index the textual content of documents in digital libraries. However, scholarly articles frequently report important findings in figures for visual impact and the contents of these figures are not indexed. These contents are often invaluable to the researcher in various fields, for the purposes of direct comparison with their own work. Therefore, searching for figures and extracting figure data are important problems. To the best of our knowledge, there exists no tool to automatically extract data from figures in digital documents. If we can extract data from these images automatically and store them in a database, an end-user can query and combine data from multiple digital documents simultaneously and efficiently. We propose a framework based on image analysis and machine learning to extract information from 2-D plot images and store them in a database. The proposed algorithm identifies a 2-D plot and extracts the axis labels, legend and the data points from the 2-D plot. We also segregate overlapping shapes that correspond to different data points. We demonstrate performance of individual algorithms, using a combination of generated and real-life images.”

paper
talk