Martin Gerlach

• PhD / Dr. rer. nat.
• Research Scientist at Wikimedia Foundation

One of the main computational and scientific challenges in the modern age is to extract useful information from unstructured texts. Topic models are one popular machine-learning approach that infers the latent topical structure of a collection of documents. Despite their success—particularly of the most widely used variant called latent Dirichlet a...

We use an information-theoretic measure of linguistic similarity to investigate the organization and evolution of scientific fields. An analysis of almost 20M papers from the past three decades reveals that the linguistic similarity is related but different from experts and citation-based classifications, leading to an improved view on the organiza...

Quantifying the similarity between symbolic sequences is a traditional problem in information theory which requires comparing the frequencies of symbols in different sequences. In numerous modern applications, ranging from DNA over music to texts, the distribution of symbol frequencies is characterized by heavy-tailed distributions (e.g., Zipf’s la...

One of the most celebrated findings in complex systems in the last decade is that different indexes y (e.g., patents) scale nonlinearly with the population~x of the cities in which they appear, i.e., $y\sim x^\beta, \beta \neq 1$. More recently, the generality of this finding has been questioned in studies using new databases and different definiti...

Zipf's law is just one out of many universal laws proposed to describe statistical regularities in language. Here we review and critically discuss how these laws can be statistically interpreted, fitted, and tested (falsified). The modern availability of large databases of written text allows for tests with an unprecedent statistical accuracy and a...

Intrinsically motivated information seeking is an expression of curiosity believed to be central to human nature. However, most curiosity research relies on small, Western convenience samples. Here, we analyze a naturalistic population of 482,760 readers using Wikipedia’s mobile app in 14 languages from 50 countries or territories. By measuring the...

Links are a fundamental part of information networks, turning isolated pieces of knowledge into a network of information that is much richer than the sum of its parts. However, adding a new link to the network is not trivial: it requires not only the identification of a suitable pair of source and target entities but also the understanding of the c...

With over 60M articles, Wikipedia has become the largest platform for open and freely accessible knowledge. While it has more than 15B monthly visits, its content is believed to be inaccessible to many readers due to the lack of readability of its text. However, previous investigations of the readability of Wikipedia have been restricted to English...

With 60M articles in more than 300 language versions, Wikipedia is the largest platform for open and freely accessible knowledge. While the available content has been growing continuously at a rate of around 200K new articles each month, very little attention has been paid to the discoverability of the content. One crucial aspect of discoverability...

Wikipedia, in its role as the world's largest encyclopedia, serves a broad range of information needs. Although previous studies have noted that Wikipedia users' information needs vary throughout the day, there is to date no large-scale, quantitative study of the underlying dynamics. The present paper fills this gap by investigating temporal regula...

With 60M articles in more than 300 language versions, Wikipedia is the largest platform for open and freely accessible knowledge. While the available content has been growing continuously at a rate of around 200K new articles each month, very little attention has been paid to the accessibility of the content. One crucial aspect of accessibility is...

Wikipedia, in its role as the world's largest encyclopedia, serves a broad range of information needs. Although previous studies have noted that Wikipedia users' information needs vary throughout the day, there is to date no large-scale, quantitative study of the underlying dynamics. The present paper fills this gap by investigating temporal regula...

Despite the importance and pervasiveness of Wikipedia as one of the largest platforms for open knowledge, surprisingly little is known about how people navigate its content when seeking information. To bridge this gap, we present the first systematic large-scale analysis of how readers browse Wikipedia. Using billions of page requests from Wikipedi...

"Wiki rabbit holes" are informally defined as navigation paths followed by Wikipedia readers that lead them to long explorations, sometimes involving unexpected articles. Although wiki rabbit holes are a popular concept in Internet culture, our current understanding of their dynamics is based on anecdotal reports only. To bridge this gap, this pape...

Every day millions of people read Wikipedia. When navigating the vast space of available topics using embedded hyperlinks, readers follow different trajectories in terms of the sequence of articles. Understanding these navigation patterns is crucial to better serve readers' needs and address structural biases and knowledge gaps. However, systematic...

Despite the importance and pervasiveness of Wikipedia as one of the largest platforms for open knowledge, surprisingly little is known about how people navigate its content when seeking information. To bridge this gap, we present the first systematic large-scale analysis of how readers browse Wikipedia. Using billions of page requests from Wikipedi...

We are interested in the widespread problem of clustering documents and finding topics in large collections of written documents in the presence of metadata and hyperlinks. To tackle the challenge of accounting for these different types of datasets, we propose a novel framework based on Multilayer Networks and Stochastic Block Models. The main inno...

We are interested in the widespread problem of clustering documents and finding topics in large collections of written documents in the presence of metadata and hyperlinks. To tackle the challenge of accounting for these different types of datasets, we propose a novel framework based on Multilayer Networks and Stochastic Block Models. The main inno...

Hyperlinks constitute the backbone of the Web; they enable user navigation, information discovery, content ranking, and many other crucial services on the Internet. In particular, hyperlinks found within Wikipedia allow the readers to navigate from one page to another to expand their knowledge on a given subject of interest or to discover a new one...

A major challenge for many analyses of Wikipedia dynamics -- e.g., imbalances in content quality, geographic differences in what content is popular, what types of articles attract more editor discussion -- is grouping the very diverse range of Wikipedia articles into coherent, consistent topics. This problem has been addressed using various approac...

Single cell RNA sequencing (scRNA-seq) data are now routinely generated in experimental practice because of their promise to enable the quantitative study of biological processes at the single cell level. However, cell type and cell state annotations remain an important computational challenge in analyzing scRNA-seq data. Here, we report on the dev...

In January 2019, prompted by the Wikimedia Movement's 2030 strategic direction, the Research team at the Wikimedia Foundation identified the need to develop a knowledge gaps index -- a composite index to support the decision makers across the Wikimedia movement by providing: a framework to encourage structured and targeted brainstorming discussions...

The use of Project Gutenberg (PG) as a text corpus has been extremely popular in statistical analysis of language for more than 25 years. However, in contrast to other major linguistic datasets of similar importance, no consensual full version of PG exists to date. In fact, most PG studies so far either consider only a small number of manually sele...

One of the most widely used approaches in natural language processing and information retrieval is the so-called bag-of-words model. A common component of such methods is the removal of uninformative words, commonly referred to as stopwords. Currently, most practitioners use manually curated stopword lists. This approach is problematic because it c...

Katahira et al. investigated the potential impact of skewness in the marginal distributions of personality trait on the findings reported by us in Gerlach et al. We concur with Katahira et al.’s finding in synthetic 2-dimensional data that there exists a mechanism by which skewness can induce detection of “meaningful clusters” using our proposed me...

The analysis of citations to scientific publications has become a tool that is used in the evaluation of a researcher’s work; especially in the face of an ever-increasing production volume1–6. Despite the acknowledged shortcomings of citation analysis and the ongoing debate on the meaning of citations7,8, citations are still primarily viewed as end...

The availability of large datasets requires an improved view on statistical laws in complex systems, such as Zipf's law of word frequencies, the Gutenberg-Richter law of earthquake magnitudes, or scale-free degree distribution in networks. In this paper we discuss how the statistical analysis of these laws are affected by correlations present in th...

The availability of large datasets requires an improved view on statistical laws in complex systems, such as Zipf's law of word frequencies, the Gutenberg-Richter law of earthquake magnitudes, or scale-free degree distribution in networks. In this paper we discuss how the statistical analysis of these laws are affected by correlations present in th...

Topic models are in widespread use in natural language processing and beyond. Here, we propose a new framework for the evaluation of probabilistic topic modeling algorithms based on synthetic corpora containing an unambiguously defined ground truth topic structure. The major innovation of our approach is the ability to quantify the agreement betwee...

The use of Project Gutenberg (PG) as a text corpus has been extremely popular in statistical analysis of language for more than 25 years. However, in contrast to other major linguistic datasets of similar importance, no consensual full version of PG exists to date. In fact, most PG studies so far either consider only a small number of manually sele...

In this Formal Comment, the authors of the recent publication "Large-scale investigation of the reasons why potentially important genes are ignored" maintain that it can be read as an opportunity to explore the unknown.

Understanding human personality has been a focus for philosophers and scientists for millennia¹. It is now widely accepted that there are about five major personality domains that describe the personality profile of an individual2,3. In contrast to personality traits, the existence of personality types remains extremely controversial⁴. Despite the...

Biomedical research has been previously reported to primarily focus on a minority of all known genes. Here, we demonstrate that these differences in attention can be explained, to a large extent, exclusively from a small set of identifiable chemical, physical, and biological properties of genes. Together with knowledge about homologous genes from m...

Study of homologous genes predicts study of human genes. (A) Prediction of the number of research publications using the model of Fig 1A, extended to include the year of the initial publications on homologous nonhuman genes (S1 Data). (B) Number of publications for individual genes conditioned on the existence of homologous genes in nonhuman model...

Mapping of PubMed IDs to Web of Science IDs. Mapping of PubMed IDs to Web of Science IDs for publications linked to genes. (XLSX)

Comparison of feature importance for prediction of the year of initial publication and the total number of publications. Median importance of features over 500 independent randomizations of the models for predicting the number of publications and the year of their discovery. (XLSX)

Nearby accessible important genes that are studied less than expected. Closest gene of S8 Table for every other gene in the 15-dimensional feature space in Fig 1B. (XLSX)

Physical, chemical, and biological features of genes predict the number of publications. (A) Ward-clustering of feature importance of 500 gradient boosting regression models. Numbers in brackets indicate order of features in heatmaps in Fig 1B. (B) Prediction of the number of publications for the 12,948 genes with a complete catalog of features usi...

Physical, chemical, and biological features mapped to individual genes. z-score of individual features for genes in the tSNE mapping of Fig 1. Numbers in brackets indicate order of features in heatmaps in Fig 1 (S1 Data). tSNE, t-distributed stochastic neighbor embedding. (TIF)

Publications reporting the discovery of new genes preferentially cite model organism. (A) As Fig 2D, but for individual years during the 1980s and 1990s, the decades in which most human genes were discovered. Also see S5D Fig (S1 Data). (B) Fraction of nonhuman organisms cited by initial publications of human genes. Enrichment represents log2 ratio...

Attention in publications closely tracks number of publications. Fractional counting, in which the occurrence of a gene in a publication counts as 1/(number of genes in publication), versus normal counting, in which the occurrence of a gene in a publication counts as 1, of publications with multiple genes (S1 Data). (TIF)

Career rewards disfavor novelty. (A) Career prospects of junior scientists correlate with the preceding attention directed towards genes: probability to transition to principal investigator (PI) status for authors of publications, according to the median attention of the genes in these publications. If, in the preceding years, this attention fell i...

Large-scale studies are a reference for many other publications. (A) Kernel-density estimation of the fraction of genes with a given number of publications versus the median number of genes co-occurring in the respective publications. The observed pattern is consistent with the notions of “small science” and “big science” (S1 Data). (B) Median perc...

Literature survey of genes with increased attention between 2011 and 2015. Enrichment in publications per gene between 2011 and 2015 over the time until 2010. The count of publications until 2010 has been normalized such that the total number of publications matches the time between 2011 and 2015. (XLSX)

Fraction of unstudied homologs. Number and fraction of unstudied homologs of unstudied human genes for different taxa. Unstudied genes were defined as in S12 Fig and marking genes that have not been covered by the research effort corresponding to a single single-gene study. (XLSX)

Accessible important genes that are studied less than expected. Genes with characteristics that have occurred in fewer publications than predicted by models of Fig 1A and carry the three favorable strategic properties described in Fig 4E (strong loss-of-function sensitivity and GWAS associations, experimental approachability, and the presence of in...

Extreme inequality in the research attention given to human protein-coding genes. (A) Frequency of the number of research publications associated with human protein-coding genes in MEDLINE. Black line shows a log-normal fit to the data (S1 Data). (B) Human-curated GO annotations for individual genes, binned by number of publications. Upper limit of...

Catalog of absence of features. (A) Hamming-clustering of genes according to absence of features (S1 Data). (B) Number of research publications for genes with and without complete catalog of features. (TIF)

Health research funding correlates with the number of publications. (A) The number of grants for genes as a function of the number of publications on a gene. (B) Correlation between the attention of NIH-sponsored research publications and the amount of allocated NIH budget on individual genes (dots). The latter is approximated by equal allocation o...

What we know about poorly studied genes. (A) Distribution of the attention (measured by fractional publications) in publications given to genes. Genes with attention levels below 1 are denoted unstudied (blue), whereas genes with attention levels above 1 are denoted studied (orange). (B) Percentage of genes with indicated characteristic. (C) As B,...

Decrease in the fraction of scientists working on model organisms. Fraction of scientists who—within the indicated year—publish exclusively on nonhuman genes (or gene products) or exclusively on human genes (or gene products), or both. The fraction of scientists who exclusively published on human genes had been stable in the 1980s and 1990s, while...

Accessible important genes. List of genes that have strong loss-of-function sensitivity and GWAS associations, experimental approachability, and the presence of invertebrate model organisms for genes in 15-dimensional feature space. GWAS, genome-wide association study. (XLSX)

Predictability of research effort. (A) Cumulative share of publications in MEDLINE covered by the fraction of most common genes in decreasing order (S1 Data). (B) Gini coefficient (a measure of inequality) for genes in publications over time. When looking at income or wealth, Gini coefficients of 0.6 are considered extreme (S1 Data). (C) Correlatio...

List of genes with an incomplete catalog of features. NCBI gene identifiers (Entrez genes), NCBI gene symbols, and Ensemble Gene IDs are provided. NCBI, National Center for Biotechnology Information. (XLSX)

Map of the 15-dimensional space. Coordinates of genes in Fig 1B. In addition, the inferred number of publications, NCBI gene symbols, and Ensemble Gene IDs are provided. NCBI, National Center for Biotechnology Information. (XLSX)

Gene-specific context for further exploration of genes. Gene-specific information to facilitate further experimentation. Tissue and cell line with highest RNA expression (“highest tissue,” “highest cells”); flag indicating whether frequently differentially expressed in EBI-GXA (https://www.ebi.ac.uk/gxa); flag indicating whether frequently reported...

One of the main computational and scientific challenges in the modern age is to extract useful information from unstructured texts. Topic models are one popular machine-learning approach which infers the latent topical structure of a collection of documents. Despite their success --- in particular of its most widely used variant called Latent Diric...

We use an information-theoretic measure of linguistic similarity to investigate the organization and evolution of scientific fields. An analysis of almost 20M papers from the past three decades reveals that the linguistic similarity is related but different from experts and citation-based classifications, leading to an improved view on the organiza...

We show how generalized Gibbs-Shannon entropies can provide new insights on the statistical properties of texts. The universal distribution of word frequencies (Zipf's law) implies that the generalized entropies, computed at the word level, are dominated by words in a specific range of frequencies. Here we show that this is the case not only for th...

We show how generalized Gibbs-Shannon entropies can provide new insights on the statistical properties of texts. The universal distribution of word frequencies (Zipf's law) implies that the generalized entropies, computed at the word level, are dominated by words in a specific range of frequencies. Here we show that this is the case not only for th...

Zipf’s law is just one out of many universal laws proposed to describe statistical regularities in language. Here we review and critically discuss how these laws can be statistically interpreted, fitted, and tested (falsified). The modern availability of large databases of written text allows for tests with an unprecedent statistical accuracy and a...

One of the most celebrated findings in complex systems in the last decade is that different indexes y (e.g., patents) scale nonlinearly with the population~x of the cities in which they appear, i.e., $y\sim x^\beta, \beta \neq 1$. More recently, the generality of this finding has been questioned in studies using new databases and different definiti...

It is well accepted that adoption of innovations are described by S-curves (slow start, accelerating period, and slow end). In this paper, we analyze how much information on the dynamics of innovation spreading can be obtained from a quantitative description of S-curves. We focus on the adoption of linguistic innovations for which detailed database...

In this paper we combine statistical analysis of large text databases and simple stochastic models to explain the appearance of scaling laws in the statistics of word frequencies. Besides the sublinear scaling of the vocabulary size with database size (Heaps' law), here we report a new scaling of the fluctuations around this average (fluctuation sc...

The concept of dominant interaction hamiltonians is introduced and applied to classical planar electron-atom scattering. Each trajectory is governed in different time intervals by two variants of a separable approximate hamiltonian. Switching between them results in exchange of energy between the two electrons. A second mechanism condenses the elec...

We propose a stochastic model for the number of different words in a given database which incorporates the dependence on the database size and historical changes. The main feature of our model is the existence of two different classes of words: (i) a finite number of core-words which have higher frequency and do not affect the probability of a new...