Citizen Science 2.0: Data Management Principles to Harness the Power of the Crowd

Abstract

Citizen science refers to voluntary participation by the general public in scientific endeavors. Although citizen science
has a long tradition, the rise of online communities and user-generated web content has the potential to greatly expand its
scope and contributions. Citizens spread across a large area will collect more information than an individual researcher can.
Because citizen scientists tend to make observations about areas they know well, data are likely to be very detailed. Although
the potential for engaging citizen scientists is extensive, there are challenges as well. In this paper we consider one such
challenge – creating an environment in which non-experts in a scientific domain can provide appropriate and accurate data
regarding their observations. We describe the problem in the context of a research project that includes the development of
a website to collect citizen-generated data on the distribution of plants and animals in a geographic region. We propose an
approach that can improve the quantity and quality of data collected in such projects by organizing data using instance-based
data structures. Potential implications of this approach are discussed and plans for future research to validate the design
are described.

Full text

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
1
Citizen Science 2.0: Data Management Principles to
Harness the Power of the Crowd
Roman Lukyanenko1 , Jeffrey Parsons , Yolanda Wiersma2,
1 Faculty of Business Administration, Memorial University, St. John’s Canada
2 Department of Biology, Memorial University, St. John’s Canada
{roman.lukyanenko, jeffreyp, ywiersma}@mun.ca
Abstract. Citizen science refers to voluntary participation by the general public
in scientific endeavors. Although citizen science has a long tradition, the rise of
online communities and user-generated web content has the potential to greatly
expand its scope and contributions. Citizens spread across a large area will
collect more information than an individual researcher can. Because citizen
scientists tend to make observations about areas they know well, data are likely
to be very detailed. Although the potential for engaging citizen scientists is
extensive, there are challenges as well. In this paper we consider one such
challenge – creating an environment in which non-experts in a scientific
domain can provide appropriate and accurate data regarding their observations.
We describe the problem in the context of a research project that includes the
development of a website to collect citizen-generated data on the distribution of
plants and animals in a geographic region. We propose an approach that can
improve the quantity and quality of data collected in such projects by
organizing data using instance-based data structures. Potential implications of
this approach are discussed and plans for future research to validate the design
are described.
Keywords: design, citizen science, management, database design, conceptual
modeling, data quality.
1 Introduction
Citizen science is a term used to describe the voluntary participation of amateur
scientists in scientific endeavors [1]. Humans are increasingly regarded as effective
sensors of their environment [2] and the potential for using information collected by
individuals is continuously expanding [3]. Citizen science has a long tradition. During
the Victorian era many wealthy individuals engaged in natural history as a hobby, and
made contributions to the understanding of species distributions and behavior as a
result. With the development of the Internet, it has become easier for ordinary people
to participate and contribute large amounts of information. Yet, given the expertise
and language gap between scientists and ordinary people, information transfer in
citizen science projects is not straightforward. While citizen scientists can offer
insights and generate new ideas [4], their lack of training and expertise results in
inconsistent and incorrect data [5,6,7]. In particular, where direct elicitation of
people’s opinions is required we can expect lower scientific accuracy of data as wider

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
audiences with lesser expertise get engaged. This research attempts to address this
problem by suggesting data management principles that maximize the quantity and
quality of information collected from non-experts.
There are many advantages to harnessing citizen scientists. Participants spread
across a large area will collect more information than an individual researcher can.
Because citizen scientists tend to make observations about areas they know well, data
are likely to be very detailed. An additional advantage is the potential longevity of
such data; some citizen science programs (e.g., the Audubon Christmas Bird Count)
have been in existence for over 100 years, resulting in data sets extending over long
periods, thus enabling analysis of trends. Coupled with the availability of relatively
inexpensive photo and video equipment, harnessing the power of ordinary people to
provide data and observations about the natural world can lead to major advances in
the natural sciences, as well as assist in vital areas of wildlife conservation and
emergency management in the event of natural disasters (such as the Gulf of Mexico
oil spill).
Although the potential for engaging citizen scientists is extensive, there are
challenges as well. In this paper we describe one such challenge – creating an online
environment in which non-experts in a scientific domain can provide appropriate and
accurate data regarding their observations. We describe the problem in the context of
a research project that includes the development of a website and database to collect
citizen-generated data on the distribution of plants and animals in a geographic
region. We propose an approach to improving the quantity and quality of data
collected in such projects by using instance-based data structures [8]. Potential
implications of this approach are discussed and plans for future research to validate
the design are described.
2 The Challenge – Facilitating Participation
The success of a citizen science project depends on the willingness and ability of
members of the general public to voluntarily observe and report information. In many
cases, this in turn requires some level of scientific knowledge by participants. For
example, the website of the Cornell Ornithology Lab, eBird (www.ebird.com ), draws
on the enthusiasm of avid birders to provide detailed information about bird sightings.
The Cornell Lab is an international leader in ornithological research, and eBird is an
exemplar of a successful online citizen science project. However, engagement of the
lay public with eBird may be limited by the application domain. Citizen scientists
who wish to upload bird sightings need to be familiar with bird taxonomy and
identification. The bird checklist provided in the online interface assumes the user has
already made a positive identification (i.e., identified the species) and knows to which
taxonomic group the bird belongs. This is acceptable for a reasonably experienced
citizen scientist, but the rank beginner ([7] provides a taxonomy of “expertise levels”
among citizen scientists) may not be able to participate, or may provide data of poor
quality as a result of his/her inability to make a positive identification [9]. Thus,
useful participation may be limited to more experienced amateurs.

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
The issue of quality and reliability of user-supplied data in citizen science projects
has attracted much attention in recent research [5]. Although the literature is limited
(given the relative recency of Web 2.0 applications), the implied assumption of much
of the work to date is that there exists an inherent trade-off between data quality and
the level of participation (data quantity). Experts are considered to be the source of
the most accurate volunteered information [7], but there are fewer “expert amateurs”
than “beginners” available to participate.
The common method of increasing data quality considered in the literature is
training and educating the volunteers. For example, data inconsistency may result
from volunteers’ lack of experience, inadequate guidelines and insufficient training
[4], “rolled up” into larger monitoring projects [6]. Training, while generally
desirable, may not always be possible, especially for low budget projects.
A typical way to increase quality is through expert verification, an approach that
has been used for by-catch and beached bird observation [12] and for unusual
observations on eBird [19]. However, with the size of data sets increasing [6],
individual verification becomes unrealistic and in many ways is contrary to the spirit
of citizen science.
Another line of research suggests social networking as key to increasing data
quality. Some research has proposed a trust and reputation model for classifying
knowledge using the social networking practice of peer evaluation of content [13].
This approach is the basis for iSpot, a website that exploits a user reputation
mechanism to determine accuracy of observations [14]. The reputation-trust model
adapted from well-developed trust research in e-commerce [e.g., 15,16,17] has been
applied to the context of citizen science [18]. While the social networking approach
appears promising, it has a number of limitations. Although it has been compared to
the “scientific peer review process” [13], social networking is useful only for popular
citizen science projects with large numbers of users. Web sites with a small number of
users may not have sufficient user activity per observation to ensure rigorous peer
review. In addition, as even a very popular website cannot guarantee that every
observation will receive equal scrutiny, this metaphor of scientific review is not fully
justified. Furthermore, users with high reputation who are considered experts in some
domain may still provide inaccurate data in other domains. Most importantly, social
networking may fail to harness the potential of an individual non-expert, as in the
absence of domain knowledge such volunteers may feel too intimidated to express
their opinion (consider the description of a type 'neophyte' [7]). Finally, the social
networking approach lacks generality, as it relies on a particular technology, and may
exclude many citizen science projects that do not currently employ a social
networking model.
Notwithstanding the value of the above approaches, we argue that it is possible to
increase the quality of data generated by of an individual volunteer by minimizing
subject information that has a high likelihood of being inaccurate. Requiring
volunteers to make a (potentially inaccurate) positive identification of natural history
phenomena implies that the observer has some knowledge of traditional scientific
taxonomy. We argue that an alternative to classifying observations according to a
fixed taxonomy is to allow volunteers to provide information about observations and
that this will increase the general success of citizen-scientist projects.

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
3 A Proposed Solution – Attribute-based Data Collection
A traditional approach to citizen participation in scientific data collection works well
(i.e., makes it possible to collect accurate data from a broad constituency) only if the
participants are capable of classifying observed phenomena accurately. For example,
accurate classification of observed plants and animals by species requires that
participants understand the distinguishing characteristics of species. We contend that
imposing this requirement on participation, as in projects such as eBird, imposes a
severe and unnecessary restriction on the level of participation that can be realized in
citizen science projects.
To combat this limitation, we propose an approach to data collection and storage
that does not require users to classify observed phenomena. Instead, they record any
attributes associated with the observation. We illustrate the approach in the context of
NLNature – an ongoing citizen scientist-based project to collect data about the flora
and fauna of Newfoundland & Labrador (www.nlnature.com ). Our proposal is based
on the instance-based data model (IBDM) [8] and our application of the model has
implications both for interface design and for database design. Working within the
framework of the IBDM, we extend the model to address issues of identifying
phenomena, and suggest how the model offers a solution to the challenges of a typical
citizen scientist project.
The IBDM is based on ontological and cognitive principles [8, 20]. Ontologically,
every “thing” possesses a unique set of properties. Classes are formed based on the
principle that one can classify things based on a subset of their observed properties,
and make inferences about unobserved properties the instance possesses by virtue of
belonging to the class [21]. Since an instance can possess very many properties, it can
belong to a very large number of potential classes, depending on the context.
By shifting the focus from a predefined classification to the thing (instance) and its
attributes (see Fig. 1) we do not need to model a domain a priori in terms of the
classes of interest. It is sufficient to ensure that the application has a comprehensive
collection of instances, and each instance contains a set of well-defined attributes.
When required, a user can assemble a dynamic classification based on the collection
of attributes that are of interest at a given moment. For example, if an attribute such as
“behavior” is of interest, then at least two classes can be constructed based on values:
animals that are nocturnal (active at night) vs. diurnal (active during the day). The
same system can also use attributes that connect each species with a biological
taxonomy to reproduce scientific biological classification. Thus, the instance-based
model is capable of achieving the objectives of a traditional classification without the
inherit limitations.

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
Fig. 1. Traditional vs. attribute-based information (Image source: Wikimedia Commons).
We posit that attribute-based design will enable potential citizen scientists to
provide data efficiently and effectively, thereby increasing their participation in data
gathering. We propose a data collection interface designed based on the primacy of a
phenomenon and its attributes over classification of the phenomenon. A user is asked
to identify those attributes (e.g., size, color, appearance, behavior, location, sound) of
an observed plant or animal. In principle, the primary scientific object of an
observation (the species observed) can be identified by an expert after the observation
is recorded, provided that the user reports enough attributes to produce a positive
identification. This contrasts with traditional approaches requiring a priori
classification (e.g., requiring users to select from a checklist of species), which are
usable only by more expert volunteers. Once several attributes are selected, the
system will match them with pre-existing sets of identifying attributes for species, and
either infer a species or ask for additional attributes that could also be automatically
inferred from those previously supplied.
Although the final attribute set resulting from an observation can potentially match
multiple species, this proposed solution offers a realistic compromise. Non-experts do
not always know the phenomenon that was observed. It is more realistic to expect a
volunteer to remember some features of unknown species then to expect a precise
classification and identification. The key activity of identification therefore shifts
from designing a perfect classification to facilitating effective attribute management.
The more the system can guide the choice of attributes, the higher inferential value
such records hold, and the easier it is to classify observations.

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
4 Attribute-based Database Design
Database structure can be either a major inhibitor or a facilitator of system evolution
[8, 20, 22]. Traditionally, database design results in a representation of the application
domain as a set of related classes (translated to tables in a relational database). In
addition, once the database structure is established it is assumed to be relatively static,
allowing other application elements, such as program code, to be created based on the
static structure. Altering the database structure once a system is built is costly. Thus,
traditional database design is subject to the inherit limitations of a rigid classification
[8].
The collection of user-supplied information based on attributes of observations
suggests the need for a database structure that supports variability in the data collected
from observers, including failure to classify an observation. Support for flexible
attribute collection can be implemented using a traditional relational database, as
illustrated in Fig. 2. We propose storing attributes in a generic table “Attributes” that
contains attribute name and a unique identifier. A separate “Attributes-Relationships”
table links one attribute to another and creates relationships between attributes. The
table contains the primary key from the parent attribute and a primary key from the
child attribute, thus making many-to-many relationships possible. For example, if the
user selects the attribute “was flying” then “lives in water” will be automatically
removed from the interface, and the system will respond by presenting a new set of
potential attributes that can be inferred from “was flying” (e.g., “has feathers”, “seen
at night”, “six legs”). The choice of the first attribute narrows the observation to a bird
(subsequent attributes could focus on feather color, beak size, habitat, etc.), the
second to a bat, and the third to a flying insect.
Fig. 2. ER diagram showing instance-based data structure for a typical citizen scientist project.
In order to match selected attributes against a class-defining set, class blueprints
for each species need to be maintained. This is achieved by a table of “Species
Definitions” that links species with their attributes via a one-to-many relationship. For
example, boreal felt lichen will link to the following attributes: fuzzy white fringe
around the edges, grayish-brown when dry, has red dots, leafy, slate-blue when moist.
User-observed properties are then matched against the class definitions to infer class
membership. If necessary, new class definitions can be added or existing ones altered
during the operational phase of the enterprise system without having to change the
database schema. Finally, we provide tables that join objects and attributes to store the

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
details of the observations. These tables store events in the system. Each table
includes primary keys from the attributes and objects tables, attribute values, and
date/time of the attribute creation/change. By recording the date of attribute
creation/change, the system can document events that happen to the same
phenomenon. This approach addresses a persistent issue of database design –
adaptation to organizational change – that a traditional approach with its reliance on
rigid classification struggles to resolve [8, 22].
5 Implications for Data Gathering
The attribute based system proposed for this citizen scientist project has the potential
to increase participation rates (and, hence, data quantity). Unlike natural history
websites that only present taxonomic checklists and assume a basic level of expertise
from citizen scientists, the system proposed here allows for the full spectrum of
volunteer contributors [7] to participate. We believe that this will provide a means of
validating user-supplied data within the user community, particularly if users supply
additional information with their observations (e.g., photographs) that can be
reviewed by experts when necessary.
Many citizen science projects provide inventory data across space and time.
Although there will be biases within the data (for example, to areas where there is
high human population density and to more charismatic or easily observable species),
the data do have the benefit of indicating long-term trends. For the scientific
community, the biggest value is that such data sets are generated by many “eyes on
the ground;” thus, there is a higher likelihood of rare or unusual species being
detected or for early detection of new trends. Hence, it is important to have a usable
system that promotes a broad and consistent level of participation. Some potential
uses of data collected this way might be unanticipated. For example, long term data
can be useful to identify benchmark conditions in the event of a natural or
anthropogenic disaster (e.g., the Gulf oil spill) and can guide restoration strategies.
This research explores general ways of facilitating information transfer between
users with different level of domain expertise within the context of a citizen science
project. Information systems are increasingly being used to collect data from ordinary
people (e.g. personal health records [23]). While a number of factors are considered to
influence information quality (e.g., [24]), little attention is given to the role of data
structures in ensuring quality of collected information.
6 Limitations and Future Research
Internet technologies open new opportunities for citizen science. Yet the knowledge
requirements implied by rigid data structures constrain effective participation of
novices and thereby limit the potential outreach of citizen science projects. A
successful implementation of the approach proposed in this paper can facilitate
development of citizen-scientist initiatives. We believe it also has broader

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
applications based on user-generated content, and promises to be a practical solution
to an important design problem in citizen science.
The foundation of our proposed approach to improving the quantity and quality of
citizen science projects is the IBDM [8]. The primary theoretical assumption of the
IBDM – that existence of things and properties (attributes) precedes classification -
has generally [cf. 25] been supported in ontological [26,27] and cognitive research
[28]. However, while attributes are building blocks of classification [29], not all
classes can be efficiently expressed as sets of common attributes (e.g., radial
categories [30,31,32]). Moreover, many superordinate categories, such as furniture,
animal, vehicles tend to be abstract and reflect some rules or functions rather than
observable attributes [33-34]. While this appears to limit the scope of our model, we
believe that for practical reasons little information in citizen science projects will be
expressed in terms of higher-level categories. Indeed, humans prefer to avoid
superordinate categories when they think of individual objects [35].
Classification is a ubiquitous activity and an attribute-centered approach to
knowledge management needs to be tested to determine its technological, economic,
scientific and business utility. We are currently designing empirical studies to
measure the practical impact of the above approach on data collection and storage,
user participation and satisfaction, data quality, and usefulness to scientists. The
experiment will also test the overall effectiveness and feasibility of applying the
IBDM to empower citizen scientists.
References
1. Silvertown, J.: A new dawn for citizen science. Trends in Ecology and Evolution 24 467-
471 (2009)
2. Goodchild, M.: Citizens as sensors: the world of volunteered geography. GeoJournal 69
211-221 (2007)
3. Hand, E.: People power. Nature 466 685-687 (2010)
4. Foster-Smith, J., Evans, S.M.: The value of marine ecological data collected by volunteers.
Biological Conservation 113 199-213 (2003)
5. Flanagin, A., Metzger, M.: The credibility of volunteered geographic information.
GeoJournal 72 137-148 (2008)
6. Wiersma, Y.F.: Birding 2.0: citizen science and effective monitoring in the Web 2.0 world.
Avian Conservation and Ecology 5 13 (2010)
7. Coleman, D.J., Georgiadou, Y., Labonte, J.: Volunteered geographic information: The
nature and motivation of producers. International Journal of Spatial Data Infrastructures
Research 4 332-358 (2009)
8. Parsons, J., Wand, Y.: Emancipating instances from the tyranny of classes in information
modeling. ACM Transactions on Database Systems 25 228 (2000)
9. Parsons, J., Lukyanenko, R., Wiersma, Y.: Easier citizen science is better. Nature 471 37
(2011)
10. Dickinson, J.L., Zuckerberg, B., Bonter, D.N.: Citizen science as an ecological research
tool: challenges and benefits. Annual Review of Ecology, Evolution, and Systematics 41
112-149 (2010)
11. Aaron, W.E.G., Tudor, M.T., Haegen, W.M.V.: The Reliability of Citizen Science: A Case
Study of Oregon White Oak Stand Surveys. Wildlife Society Bulletin 34 1425-1429 (2006)

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
12. Hamel, N.J., Burger, A.E., Charleton, K., Davidson, P., Lee, S., Bertram, D.F., Parrish,
J.K.: Bycatch and beached birds: Assessing mortality impacts in coastal net fisheries using
marine bird strandings. Marine Ornithology (2009)
13. Bishr, M., Mantelas, L.: A trust and reputation model for filtering and classifying
knowledge about urban growth. GeoJournal 72 229-237 (2008)
14. Silvertown, J.: Taxonomy: include social networking. Nature 467 788-788 (2010)
15. Komiak, S.Y.X., Benbasat, I.: The effects of personalization and familiarity on trust and
adoption of recommendation agents. MIS Quarterly 30 941-960 (2006)
16. Gefen, D., Karahanna, E., Straub, D.W.: Trust and TAM in online shopping: An integrated
model. MIS Quarterly 27 51-90 (2003)
17. Palvia, P.: The role of trust in e-commerce relational exchange: A unified model.
Information & Management 46 213-220 (2009)
18. Alabri, A., Hunter, J.: Enhancing the quality and trust of citizen science data. IEEE
eScience 2010, Brisbane, Australia (2010)
19. Sullivan, B.L., Wood, C.L., Iliff, M.J., Bonney, R.E., Fink, D., Kelling, S.: eBird: A
citizen-based bird observation network in the biological sciences. Biological Conservation
142 2282-2292 (2009)
20. Parsons, J., Su, J.: Analysis of data structures to support the instance-based data model.
International Conference on Design Science Research in Information Systems and
Technology (DESRIST) (2006) 107-130
21. Parsons, J., Wand, Y.: A question of class. Nature 455 1040-1041 (2008)
22. Allen, B.R., Boynton, A.C.: Information architecture: In search of efficient flexibility. MIS
Quarterly 15 435-445 (1991)
23. Agarwal, R., Angst, C.M.: Technology-Enabled Transformations in U.S. Health Care:
Early Findings on Personal Health Records and Individual Use. In: Galletta, D., Zhang, P.
(eds.): Human-Computer Interaction and Management Information Systems: Applications.
M.E. Sharpe, Inc, Armonk, NY (2006)
24. Nicolaou, A.I., McKnight, D.H.: Perceived information quality in data exchanges: Effects
on risk, trust, and intention to use. Information Systems Research 17 332-351 (2006)
25. Grill-Spector, K., Kanwisher, N.: Visual recognition. Psychological Science 16 152-160
(2005)
26. Bunge, M.A.: Treatise on Basic Philosophy: The furniture of the world. Reidel, Dordrecht ;
Boston (1977)
27. Wand, Y., Weber, R.: An ontological model of an information system. IEEE Transactions
on Software Engineering 16 1282-1292 (1990)
28. Bowers, J.S., Jones, K.W.: Detecting objects is easier than categorizing them. Quarterly
Journal of Experimental Psychology 61 552-557 (2008)
29. Wand, Y., Monarchi, D.E., Parsons, J., Woo, C.C.: Theoretical foundations for conceptual
modeling in information systems development. Decision Support Systems 15 285-304
(1995)
30. Raccoon, L.S.B., Puppydog, P.O.P.: A middle-out concept of hierarchy (or the problem of
feeding the animals). SIGSOFT Softw. Eng. Notes 23 111-119 (1998)
31. Young, J.J., Williams, P.F.: Sorting and comparing: Standard-setting and "ethical"
categories. Critical Perspectives on Accounting 21 509-521 (2010)
32. Lakoff, G.: Women, fire, and dangerous things : what categories reveal about the mind.
University of Chicago Press, Chicago (1987)
33. Rosch, E., Mervis, C.B., Gray, W.D., Johnson, D.M., Boyesbraem, P.: Basic objects in
natural categories. Cognitive Psychology 8 382-439 (1976)
34. Murphy, G.L., Wisniewski, E.J.: Categorizing objects in isolation and in scenes - What a
superordinate is good for. Journal of Experimental Psychology: Learning 15 572-586
(1989)

Published under License Creative Commons Attribution Non-commercial. Citation: Lukyanenko
R, J Parsons, YF Wiersma. 2011. Citizen science 2.0: data management principles to harness
the power of the crowd. Lecture Notes in Computer Science 6629:465-473.
35. Rorissa, A.: User-generated descriptions of individual images versus labels of groups of
images: A comparison using basic level theory. Information Processing & Management 44
1741-1753 (2008)