Apappu writeup on Lee Giles

• Author asserts that each discipline of science has a cynosure object like an algorithm for CS, compound structure for Chemistry, etc. Therefore, scalable and independent extraction techniques have to be employed to maintain index of such objects.

• The new CiteseerX could provide information about tables, figures and authors (including disambiguation) of papers. This seems to be a natural extension to its predecessor

which was good at pulling out citations and author information from the indexed papers. Author states that 3D graphical figures are tough nuts to crack and they are still working on it.

• A little window has been opened into the system while he discussed about using co-author, affiliation, email, address features to disambiguate names. A little more information was

given about using hierarchical CRFs for name/formulae tagging/segmenting.

• I think segmentation of sub-structures in chemical IUPAC names is an interesting problem that was mentioned during the talk. This is more or less a morphological analysis problem.

• I liked the citation-recommendation feature and I am looking forward to use it.