Bias, may it be in sampling or judgment, is not a new topic. However, with the increasing usage of data and models trained from them in almost all areas of everyday life, the topic rapidly gains relevance to the broad public. Even more, the opportunistic reuse of data (traces) that characterizes today's data science calls for new ways to understand and mitigate the e ects of biases. Here, we discuss biases in the context of Linked Data, ontologies, and reasoning services and point to the need for both technical and social solutions. We believe that debiasing knowledge graphs will become a pressing issue as these graphs enter everyday life rapidly. This is a provocative topic, not only from a technical perspective but because it will force us as a Semantic Web community to discuss whether we want to debias in the rst place and who gets a say in how to do so.
The bias-variance dilemma describes the struggle of trying to minimize the two main sources of error that plague machine learning algorithms in their attempt to generalize beyond their training data. While this trade-o relation is well known, it has only recently reached broad public attention due to the mainstream adoption of machine learning methods in everyday electronics and spectacular failures of their learned models such as categorizing black people as gorillas or auto-suggesting to kill all Jews in typeahead search.
While these cases generate the most public outcry, a di erent type of error
may have subtle but more serious long-term consequences, namely
representational bias such as Google's image search for \CEO" depicting mostly males.
Even if this might still re ect today's reality and history, we would not want
an intelligent system to learn this and consequently, for instance, recommend
doctor as a career choice for men and nurse for women [
There are ongoing discussions within the machine learning community about how to address representational biases (among other kinds), but these discussions have not yet reached the Knowledge Graph and Semantic Web communities despite representational issues being at their core. We believe that the heterogeneity of the Linked Data cloud, i.e., its decentralized nature and contributions from many sources and cultures, can o er protections against biases to a certain degree, but this will come at a cost of increased variances.
Biases in knowledge graphs (KGs), as well as potential means to address them, di er from those in linguistic models or image classi cation. Instead of learning the meaning of a term by observing the context in which it arises within a large corpus or classifying buildings from millions of labeled images, KGs are sparse in the sense that only a small number of triples are available per entity. These triples are statements about the world, not usage patterns. Finally, it's important to recognize the term bias has di erent meanings across communities, implying that models can be unbiased in a machine learning sense yet simultaneously show substantial bias in a cultural context, e.g., when the training data do not re ect evolving social consensus; we address the latter kind of bias here.
In this vision paper, we describe biases from three di erent perspectives,
namely those arising from the available data, those embedded in ontologies (i.e.,
the schema level), and nally those that are a result of drawing inferences. We
illustrate each type with a small experiment.
The sheer size of the Linked Data cloud may lead to the impression that it is
safe from selection biases such as sampling bias. However, the data available to
both open and commercial knowledge graphs today is, in fact, highly biased.
Coverage serves as an illustrative example of the underlying problem. Figure 1
plots more than one million geographically-located entities from DBpedia (en) in
red. These entities, most of them being places such as Ford's Theater, link actors
such as Abraham Lincoln, to events such as his assassination, to objects such as
Booth's weapon. Consequently, the global coverage of places reveals much about
the underlying distribution of non-spatial data as well. The map also shows
Landscan-based estimates of population density. As can be seen, coverage is not
uniform. For Europe, Japan, Australia, and the US, marker density strongly
correlates with population density, whereas this trend breaks for large parts of
Asia, Africa, and South America. At rst, one may expect that these results are
re ections of using the English DBpedia version, however, the resulting pattern
largely remains the same when comparing other language versions and data
sources such as GeoNames, social media postings, or even government data [
Put more provocatively, we know a lot about the western world and we do so from a western perspective. Even more, most of what we know about other regions and cultures comes from this same western perspective. This is an artifact of history, di erences in cultures of data sharing, availability of free governmental data, nancial resources, and so forth. The e ects are similar to what has been discussed in the machine learning community with respect to biases in word embeddings or in image search and tagging. This is not a minor issue. It translates into biased knowledge graph embeddings that increase dissimilarity to those less prototypical cases, it in uences recommender system and question answering, and it learns rules from biased training data.
As a substantial part of the Linked Data technology stack is based on open standards and is available as open source implementations, we hope that a more diverse set of contributors will form around it, thereby making the Linked Data cloud more robust to biases, nancial incentives, and so forth. 3
In addition to data bias, knowledge graphs encounter another type of bias at
the schema/ontologies level. Most ontologies are developed in a top-down
manner with application needs in mind, or in case of top-level ontologies, certain
philosophical stances. They are typically de ned by a group of engineers in
collaboration with domain experts, and thus also implicitly re ect the worldviews
and biases of the development team. Such ontologies will likely contain most of
the well-known human biases and heuristics such as anthropocentric thinking.
The increasing use of bottom-up techniques such as machine learning to derive
axioms/rules from data will not mitigate these problems as the resulting models
will fall victim to the data biases discussed before. In addition, ontologies may
be a ected by other biases such as so-called encoding bias [
Anthropocentric thinking and application needs are, for instance, at play in many of the vocabularies used to describe Points Of Interest. They typically contain dozens or even hundreds of classes for various sub-classes of restaurants, bars, and music venues, but only a handful of classes for natural features such as rivers. Consider, for example, the Place branch of the DBpedia ontology (2015); although it contains 168 classes, the average in- and out-degree of each class is only 0.994 when the class hierarchy is visualized as a network. The average path length is 2.18, suggesting that it is relatively shallow and at. By running
PageRank with a teleport probability of 0.85, more than 75% of the classes have a near zero PageRank score, meaning that much of its expressivity is occupied by less than 25% of classes. Simply put, the relative importance of classes is unevenly distributed in the ontology despite its general purpose characteristics.
It is worth noting that many biases are not directly encoded in the ontology
but only become visible when comparing multiple ontologies together with their
respective datasets. For example, DBpedia, GeoNames, and the Getty Thesaurus
of Geographic Names (TGN) all contain a Theater class. From a data-driven
perspective, one may assume that computing (spatial) statistics for all members of
this class, such as intensity, interaction, and point patterns, would yield similar
results across datasets [
We believe that the multitude of ontologies developed for the Semantic Web are a strength rather than a weakness as their diversity may help to mitigate some of the issues outlined above. 4
Another potential source of bias arises at the inferencing level, such as reasoning, querying, or rule learning. To start with a simple but easily overlooked example, the results of a SPARQL query depend on the entailment regimes (e.g., simple vs. RDFS entailment). However, the con guration of a particular query endpoint is outside the control of a data creator or ontology engineer and thus one may nd that multiple endpoints utilizing the same ontologies and datasets, e.g., local copies of DBpedia, yield di erent results for the same SPARQL query.
Here we focus on another aspect, namely, the relation of learning a
(correct) model that collides with social consensus. Consider for example, the use
of association rule mining to infer new rules from knowledge graphs[
In this vision paper, we highlighted the need to establish debiasing knowledge
graphs as a novel research theme for the Semantic Web / Knowledge Graph
community that di ers from current mainstream research. We highlighted multiple
sources of bias using small experiments. We believe that the topic is
provocative as it walks the ne line between social responsibility and censorship. Risk
mainly arises from the fact that debiasing itself is not a neutral task but based
on social norms that may di er by countries. Will we develop methods that can
be used for censorship and manipulation? As far as the time horizon is
concerned, we believe that this will become an equally pressing issue for the SW
community as it is currently in machine learning and that it should be openly
addressed in workshops or panels. Finally, the topic may also be approached
from a Web Science [