Page 269 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2015 Vol. 41

268 Journal of Library Science in China, Vol. 7, 2015

parameter b as about 0.29 based on the exponential model of citation distribution. Citation rate (C/
N), weighted Pt 1 citation amount, weighted Pt 2 citation amount, weighted Pt 1 citation rate (CW 1 /
N), weighted Pt 2 citation rate (CW 2 /N), h index, P index, Pt 1 index and Pt 2 index are all calculated.
Also calculated are some correlation coefficients such as between papers number, citations number,
citation rate and other indices, weighted Pt 1 citations amount, weighted Pt 1 citation rate, Pt 1 index
and other indices, weighted Pt 2 citations amount, weighted Pt 2 citation rate, Pt 1 index and other
indices.
It is clarified by the research results that compared with h-index, the two Pt indices, with higher
discrimination and sensitivity, are effective and comprehensive evaluation index of quality factor
and quantity factor; compared with P-index, Pt 1 -index has the potentiality to identify would-be
high-value literature with short publication age, and Pt 2 index has the ability to identify classic
high-value literature. There is still much to improve in this paper such as the necessity of the
further test of the conclusion out of the one-sidedness of the concerned sample discipline and the
possibly increasing difficulty of calculation as a result of the modified formula which is more
complex than the original one.

Library resource recommendation based on analysis on newswires

Junpeng CHEN & Wei YU 1 ∗

With the development of the Web, reading is more regarded as a kind of entertainment such as
reading twitter or blog than the study with in-depth thoughts. The real-time news, for example,
is a kind of popular Web-published information which can help people catch the update news
from the Web. At the same time, the works in library which contain in-depth thoughts and domain
knowledge are often overlooked in daily life.
There is a lack of research for providing professional domain knowledge and extending reading
list to users who are interested in the special topics mentioned in the real-time newswires.
Meanwhile, there is a large scale of domain knowledge and application examples in library
collections which can help users have a good understanding for those special topics. Hence, in
this paper, we provide a novel method to link the corresponding real-time news and records in the
library. The extending reading list from the library can be recommended with the technology of
natural language processing and semantic analysis.
We recommend the related library records to users who are interested in the target news. We
adopt natural language processing technology and LSA, NMF, and WMF methods to carry out our
experiments.
We use the catalogue records corpora: WorldCat-million dataset released by the OCLC in 2012.

* Correspondence should be addressed to Wei YU, Email: yuw.nju@gmail.com, ORCID: 0000-0003-1933-5380