Limits...
A Bibliometric Analysis on Cancer Population Science with Topic Modeling.

Li DC, Rastegar-Mojarad M, Okamoto J, Liu H, Leichow S - AMIA Jt Summits Transl Sci Proc (2015)

Bottom Line: We applied two topic modeling techniques: author topic modeling (AT) and dynamic topic modeling (DTM).Our initial results show that AT can address reasonably the issues related to investigators' research interests, research topic distributions and popularities.In compensation, DTM can address the evolving trend of each topic by displaying the proportion changes of key words, which is consistent with the changes of MeSH headings.

View Article: PubMed Central - PubMed

Affiliation: Biomedical Statistics & Informatics, Mayo Clinic, Rochester, MN.

ABSTRACT
Bibliometric analysis is a research method used in library and information science to evaluate research performance. It applies quantitative and statistical analyses to describe patterns observed in a set of publications and can help identify previous, current, and future research trends or focus. To better guide our institutional strategic plan in cancer population science, we conducted bibliometric analysis on publications of investigators currently funded by either Division of Cancer Preventions (DCP) or Division of Cancer Control and Population Science (DCCPS) at National Cancer Institute. We applied two topic modeling techniques: author topic modeling (AT) and dynamic topic modeling (DTM). Our initial results show that AT can address reasonably the issues related to investigators' research interests, research topic distributions and popularities. In compensation, DTM can address the evolving trend of each topic by displaying the proportion changes of key words, which is consistent with the changes of MeSH headings.

No MeSH data available.


Related in: MedlinePlus

Topic Proportion of DCP Publications
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4525228&req=5

f2-2091979: Topic Proportion of DCP Publications

Mentions: Figure 2 and Figure 3 show the ordered proportion of the 20 topics for DCP publications and for DCCPS publications respectively. In order to find out is defined as the number what each topic was focused on, we assigned each topic a name based on the words with posterior probabilities higher than some threshold and also assigned a number to refer it. For DCP publications, most of the 20 topics involved specific cancer preventions while the top five focused on studying cancer mechanisms from genomic source. It looks that modern cancer studies attempt to understand the internal causes of pathological changes from biological structures. Topics of both DCCPS and DCP publications involve breast cancer, colorectal cancer and ovarian cancer. But obviously, there are not so many specific cancer studies in DCCPS as in DCP (more than half). It is understandable since research of DCP aims at solving problems of cancer preventions from disease itself while that of DCCPS at study of characteristics of cancer population and epidemiological control of cancer spreading. Thus, topics related to statistical studies (Topic 1, 16, 13 and 20 for example) can be found in DCCPS. In the following section, we will look into the details of each topic to get a more fine level of understanding the relationship between author, topic and key words.


A Bibliometric Analysis on Cancer Population Science with Topic Modeling.

Li DC, Rastegar-Mojarad M, Okamoto J, Liu H, Leichow S - AMIA Jt Summits Transl Sci Proc (2015)

Topic Proportion of DCP Publications
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4525228&req=5

f2-2091979: Topic Proportion of DCP Publications
Mentions: Figure 2 and Figure 3 show the ordered proportion of the 20 topics for DCP publications and for DCCPS publications respectively. In order to find out is defined as the number what each topic was focused on, we assigned each topic a name based on the words with posterior probabilities higher than some threshold and also assigned a number to refer it. For DCP publications, most of the 20 topics involved specific cancer preventions while the top five focused on studying cancer mechanisms from genomic source. It looks that modern cancer studies attempt to understand the internal causes of pathological changes from biological structures. Topics of both DCCPS and DCP publications involve breast cancer, colorectal cancer and ovarian cancer. But obviously, there are not so many specific cancer studies in DCCPS as in DCP (more than half). It is understandable since research of DCP aims at solving problems of cancer preventions from disease itself while that of DCCPS at study of characteristics of cancer population and epidemiological control of cancer spreading. Thus, topics related to statistical studies (Topic 1, 16, 13 and 20 for example) can be found in DCCPS. In the following section, we will look into the details of each topic to get a more fine level of understanding the relationship between author, topic and key words.

Bottom Line: We applied two topic modeling techniques: author topic modeling (AT) and dynamic topic modeling (DTM).Our initial results show that AT can address reasonably the issues related to investigators' research interests, research topic distributions and popularities.In compensation, DTM can address the evolving trend of each topic by displaying the proportion changes of key words, which is consistent with the changes of MeSH headings.

View Article: PubMed Central - PubMed

Affiliation: Biomedical Statistics & Informatics, Mayo Clinic, Rochester, MN.

ABSTRACT
Bibliometric analysis is a research method used in library and information science to evaluate research performance. It applies quantitative and statistical analyses to describe patterns observed in a set of publications and can help identify previous, current, and future research trends or focus. To better guide our institutional strategic plan in cancer population science, we conducted bibliometric analysis on publications of investigators currently funded by either Division of Cancer Preventions (DCP) or Division of Cancer Control and Population Science (DCCPS) at National Cancer Institute. We applied two topic modeling techniques: author topic modeling (AT) and dynamic topic modeling (DTM). Our initial results show that AT can address reasonably the issues related to investigators' research interests, research topic distributions and popularities. In compensation, DTM can address the evolving trend of each topic by displaying the proportion changes of key words, which is consistent with the changes of MeSH headings.

No MeSH data available.


Related in: MedlinePlus