High-Reproducibility and High-Accuracy Method for Automated Topic Classification

Lancichinetti, Andrea; Sirer, M. Irmak; Wang, Jane X.; Acuña, Daniel E.; Körding, Konrad P.; Amaral, Luı́s A. Nunes

doi:10.1103/physrevx.5.011007

Cited by 76 publications

(91 citation statements)

References 33 publications

(39 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…To overcome this limitation of the journal-level analysis, we must determine the research topic of each publication at a finer scale. To this end, we use a highly accurate and reproducible topic classification algorithm to identify the topics of publications [35]. We identify a total of 69 topics using the titles and abstracts from the set of 61,116 publications by molecular biology faculty in our database.…”

Section: Resultsmentioning

confidence: 99%

“… Topic represents the topic number identified by the topic classification algorithm and is field-specific [35]; Outlier topic represents the topic in Fig 5.…”

Section: Resultsmentioning

confidence: 99%

“…The topic numbers and words are given by the topic classifying method [35], and the journals are those in which the number of publications is significantly more than expected to occur by chance if drawn from a hypergeometric distribution.…”

Section: Supporting Informationmentioning

confidence: 99%

See 2 more Smart Citations

Differences in Collaboration Patterns across Discipline, Career Stage, and Gender

et al. 2016

Self Cite

View full text Add to dashboard Cite

Collaboration plays an increasingly important role in promoting research productivity and impact. What remains unclear is whether female and male researchers in science, technology, engineering, and mathematical (STEM) disciplines differ in their collaboration propensity. Here, we report on an empirical analysis of the complete publication records of 3,980 faculty members in six STEM disciplines at select U.S. research universities. We find that female faculty have significantly fewer distinct co-authors over their careers than males, but that this difference can be fully accounted for by females’ lower publication rate and shorter career lengths. Next, we find that female scientists have a lower probability of repeating previous co-authors than males, an intriguing result because prior research shows that teams involving new collaborations produce work with higher impact. Finally, we find evidence for gender segregation in some sub-disciplines in molecular biology, in particular in genomics where we find female faculty to be clearly under-represented.

show abstract

Section: Resultsmentioning

confidence: 99%

“… Topic represents the topic number identified by the topic classification algorithm and is field-specific [35]; Outlier topic represents the topic in Fig 5.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Differences in Collaboration Patterns across Discipline, Career Stage, and Gender

et al. 2016

Self Cite

View full text Add to dashboard Cite

show abstract

“…In another work, Tumminello et al (2011) used the hypergeometric distribution and measured the P-value for each subset of the bipartite network. Moreover, Lancichinetti et al (2015) proposed a community detection method to classify topics to articles more efficiently, and Serrano et al (2009) used a disparity filtering method to infer significant weights in networks. Finally, Ronen et al (2014) adopted a statistical approach to determine significant links between languages in various written documents.…”

Section: Methodsmentioning

confidence: 99%

Mapping bilateral information interests using the activity of Wikipedia editors

et al. 2015

Self Cite

View full text Add to dashboard Cite

We live in a global village where electronic communication has eliminated the geographical barriers of information exchange. The road is now open to worldwide convergence of information interests, shared values and understanding. Nevertheless, interests still vary between countries around the world. This raises important questions about what today's world map of information interests actually looks like and what factors cause the barriers of information exchange between countries. To quantitatively construct a world map of information interests, we devise a scalable statistical model that identifies countries with similar information interests and measures the countries' bilateral similarities. From the similarities we connect countries in a global network and find that countries can be mapped into 18 clusters with similar information interests. Through regression we find that language and religion best explain the strength of the bilateral ties and formation of clusters. Our findings provide a quantitative basis for further studies to better understand the complex interplay between shared interests and conflict on a global scale. The methodology can also be extended to track changes over time and capture important trends in global information exchange.

show abstract

“…Due to the high degeneracy of the likelihood landscape, standard optimization algorithms will more likely infer different models after different optimization runs than infer the model with the highest likelihood,as has been previously reported Blei et al (2010a); Wallach et al (2009). A research on the validity of LDA optimization algorithms for inferring topic models also proposed that current implementations of LDA had low validity (Lancichinetti et al, 2015).…”

Section: Introductionmentioning

confidence: 99%

Hierarchical Latent Semantic Mapping for Automated Topic Generation

Zhou¹,

Chen²

2016

IJNDC

View full text Add to dashboard Cite

Much of information sits in an unprecedented amount of text data. Managing allocation of these large scale text data is an important problem for many areas. Topic modeling performs well in this problem. The traditional generative models (PLSA,LDA) are the state-of-the-art approaches in topic modeling and most recent research on topic generation has been focusing on improving or extending these models. However, results of traditional generative models are sensitive to the number of topics K, which must be specified manually and determines the rank of solution space for topic generation. The problem of generating topics from corpus resembles community detection in networks. Many effective algorithms can automatically detect communities from networks without a manually specified number of the communities. Inspired by these algorithms, in this paper, we propose a novel method named Hierarchical Latent Semantic Mapping (HLSM), which automatically generates topics from corpus. HLSM calculates the association between each pair of words in the latent topic space, then constructs a unipartite network of words with this association and hierarchically generates topics from this network. We apply HLSM to several document collections and the experimental comparisons against several state-of-the-art approaches demonstrate the promising performance.

show abstract

High-Reproducibility and High-Accuracy Method for Automated Topic Classification

Cited by 76 publications

References 33 publications

Differences in Collaboration Patterns across Discipline, Career Stage, and Gender

Differences in Collaboration Patterns across Discipline, Career Stage, and Gender

Mapping bilateral information interests using the activity of Wikipedia editors

Hierarchical Latent Semantic Mapping for Automated Topic Generation

Contact Info

Product

Resources

About