=Paper=
{{Paper
|id=Vol-1960/paper6
|storemode=property
|title=A Bayesian Framework for Reputation in Citizen Science
|pdfUrl=https://ceur-ws.org/Vol-1960/paper6.pdf
|volume=Vol-1960
|authors=Joan Garriga,Jaume Piera,Frederic Bartumeus
}}
==A Bayesian Framework for Reputation in Citizen Science==
https://ceur-ws.org/Vol-1960/paper6.pdf
A Bayesian Framework for
Reputation in Citizen Science
Joan Garriga1,2 , Jaume Piera1,3 , and Frederic Bartumeus1,2,4
1
Centre de Recerca Ecològica i Aplicacions Forestals (CREAF),
Carrer de les Cases Sert 54, 08193, Cerdanyola del Vallès, Barcelona
2
Centre d’Estudis Avançats de Blanes (CEAB-CSIC),
Carrer Accés Cala Sant Francesc 14, 17300, Girona
3
Institut de Ciències del Mar (ICM-CSIC),
Passeig Marı́tim de la Barceloneta 37-49, 08003, Barcelona
4
Institució Catalana de Recerca i Estudis Avançats (ICREA),
Passeig de Lluı́s Companys 23, 08010, Barcelona
jgarriga@ceab.csic.es
Abstract. The viability of any Citizen Science (CS) research program
is absolutely conditioned to the engagement of the citizen. In a CS frame-
work in which participants are expected to perform actions that can be
later on validated, the incorporation of a reputation system can be a
successful strategy to increase the overall data quality and the likelihood
of engagement, and also to evaluate how close citizens fulfill the goals of
the CS research program. Under the assumption that participant actions
are validated using a simple discrete rating system, current reputation
models, thoroughly applied in e-platform services, can be easily adapted
to be used in CS frameworks. However, current reputation models im-
plicitly assume that rated items and scored agents are the same entity,
and this does not necessarily hold in a CS framework, where one may
want to rate actions but score the participants generating it. We present
a simple approach based on a Bayesian network representing the flow
described above (user, action, validation), where participants are aggre-
gated in a discrete set of user classes and we use the global evidence in the
data base to estimate both the prior and the posterior distribution of the
user classes. Afterwards, we evaluate the expertise of each participant by
computing the user-class likelihood of the sequence of actions/validations
observed for that user. As a proof of concept we implement our model
in a real CS case, namely the Mosquito Alert project.
Keywords: citizen science, reputation system, Bayesian network
1 Introduction
Since its origins, back in the mid-90’s, citizen science (CS) has been questioned
by the scientific community as an adequate scientific methodology [7]. Pros and
cons aside, a basic principle to bring citizens and scientists into a productive
relationship is to match the public understanding of science with the science’s
1
understanding of the public [7]. To this end, modern citizen science is rethinking
methods for citizen engagement [3, 1]. Key concepts in participants engagement
are connection and reward. Connection refers to connecting the scientific goals
of the CS research program with the citizen perception of a social worry or
interest (the basic motivation to start cooperating). Reward refers to providing
feedback that can be neatly perceived as a reward (the basic motivation to keep
cooperating). Nevertheless, it is well known by psychologists [1] that the effect
of a reward resides in its expectation and vanishes as soon as it is achieved.
Thus, in order to increase the likelihood of participation in the long run, it
is necessary to generate continuous reward expectations. A successful strategy
to achieve sustained participation requires the implementation of a reputation
system as a core component of any CS research program. In addition, well-
grounded reputation systems provide back-end information of participants that
is valuable to augment data quality and to increase the fitness for use [12] of CS
research programs.
Reputation is a broad concept not only suitable to people but also to many
kinds of things or services [2]. Extending the notion given in [11], reputation
is the perception that an agent (or item) creates through past actions about its
intentions, norms, knowledge, expertise or value. Reputation can be seen as an
asset, not only to promote oneself, but also to help us to make sound judgments
in the absence of any better information. However, reputation is highly contex-
tual and what works well in a specific context may inevitably fail in many others.
As a consequence details about reputation systems are profusely treated in the
literature [2, 8, 6, 11, 14]. The simplest reputation systems scale down to a rank-
ing/voting system where information is aggregated into a single score used to
qualify and sort items (e.g. songs in iTunes, users in Stackoverflow ). Systems for
collecting and representing reputation information are usually based on simple
rating mechanisms such as thumbs-up/down or a five star rating. The difficulties
arise at the time of aggregating this information.
Many rating aggregation systems recently proposed (e.g. Amazon, iTunes,
YouTube) are different forms of Bayesian Rating (BR), a pseudo-average com-
puted as a weighted combination of the average rating of a particular item and
the average rating for all items. In a k-way rating system, (i.e with k discrete
Pmr ∈ {1, . . . , k}), with a total of m rates and an overall rating
rating levels
r̄ (all) = j=1 r (y)j /m, the BR of an item y with n ratings, and an average
Pn
rating r̄ (y) = j=1 r (y)j /n is given by,
n r̄ (y) + m r̄ (all)
BR (y) = = w r̄ (y) + (1 − w) r̄ (all) (1)
m+n
with w = n/ (n + m). A clear benefit of BR is that an item with only a few
ratings (i.e w → 0) will approach the overall mean rating, hence, does not
receive the lowest (unfair and discouraging) rate but the average rate, while the
more the item is rated (i.e n 0) the largest the weight of its own average
rating. In any case m n, hence the scoring is focused on the quality of the
ratings rather than on the quantity of ratings.
2
The Beta reputation system [9] (binomial) and the Dirichlet reputation sys-
tem [10] (DR), the multinomial generalization of the former, are reputation mod-
els based on a sound statistical machinery that explains away the Bayesian rating
concept and frames it in a real Bayesian perspective. Consider a k-way rating
system and let the rating level be indexed by i (i.e. 1 ≤ i ≤ k). Let n (y)i be
the rating counts for item y (the observed evidence), and let ai be a base rate
expressing the biases in our prior belief about each rating level. A Dirichlet (or
Beta for k = 2) rating yields a multinomial probability distribution S (y)i over
the k rating levels, where the expectation value for each rating level is computed
as,
n (y)i + C ai
S (y)i = (2)
C + n (y)
Pk
where n (y) = i=1 n (y)i and C is the a priori constant that can be set to
C = k if we consider a uniform prior (i.e. ai = 1/k). In this case Equation 2
defaults to the classical Laplace smoothing [9]. The larger the value of C with
respect to k the less the influence of the observed ratings and the more S (y)i
will approach the base rate ai . Assuming the k rating levels evenly distributed
in the range [0, 1], a point estimate reputation score is computed as,
k
X i
DR (y) = S (y)i (3)
i=1
k
Multinomial form aside, the similarity with BR (Eq. 1) is clear. But the
difference can not be obviated: while the weighting in Equation 1 emerges from
a pure frequentist persepective, in a DR the factor C ai can convey specific a
priori information provided by domain experts or any other external source [9].
Agents (and in particular human agents) may change their behaviour over
time. This issue is usually approached either by incorporating a cutoff factor
that limits the series of ratings to the most recent ones, up to a given period or
a given number, or by introducing a longevity factor that assigns a time relative
weight to ratings [9]. An additional concern in reputation systems for e-service
platforms is its resistance against typical strategies for reputation cheating (e.g.
whitewashing, sybil attacks, fraudulent ranking) reviewed in e.g. [4].
CS research programs constitute a different scenario where the aim is not
to promote user interaction but to collect useful data for their scientific goals.
Hence, reputation issues do not arise from peer-to-peer interaction but from the
need to increase citizen engagement and data quality. However, a systematic
review of 234 CS research programs presented in [5] reveals that despite of this
general concern on data quality, very few has been made in terms of participants’
reputation. Data validation is usually performed by a core of domain experts or
project coordinators, eventually assited by authomated methods or with some
level of intra-community interaction (e.g. eBird, Galaxy Zoo, iSpot) or more
broadly via crowd-sourcing (e.g. www.crowdcrafting.org). In a few cases, local
coordinators take into account the participants’ experience for validating data
3
(e.g Common Bird Monitoring, Weather Observations Website), and just in a
handful of them it is the community of participants itself who directly validate
data (e.g Galaxy Zoo, iSpot, oldWeather ). Among the later, Notes from Nature
and iSpot [15] are the ones going further in terms of community-based validation
and participants’ reputation, implementing simple agreement based algorithms
to rank participants and assign digital badges in recognition of specific achieve-
ments. Community-based validation explicitly requires a core reputation system
integrated with the CS research program. However, there is not a general ap-
proach and each research program implements reputation in a functional way to
fit its needs, neither framing its system in general conceptual frameworks, nor
making it available to the scientific community.
Notably, an implicit assumption in any of the reputation models above is that
the agent (or item) being rated is the one that is scored and, more explicitely,
that the rating system used to collect ratings for an agent is the rating system
used to score that agent (e.g. Equation2). This apparently obvious and irrel-
evant assumption might become more subtle in a CS framework. CS research
programs expect participants to perform a set of actions (basically, reporting
information in specific formats) and these actions are later on validated. In this
case, the rated items are the actions, but the scored agents are the participants.
Importantly, each kind of action might require its own discrete rating system
(not necessarily coincident in the number of levels). Yet the expertise of partic-
ipants might be expressed based on a specific set of user classes (with its own
number of levels), and scored based on the ratings of all their possibly different
actions. A straightforward way to overcome this problem is to compute separate
scores for each type of action and get an overall score using a weighted combina-
tion of the former. Alternatively, we propose a novel model for user reputation
based on a Bayesian network describing the characteristic flow of CS research
programs, (i.e. user, action, validation). The proposed method (i) decouples ac-
tion rating from participant scoring, (ii) provides a unified framework to process
validation information regardless of the rating levels used for each kind of action,
(iii) accounts for a good balance of both quality and quantity of evidence, and
(iv) is more responsive to participants’ actions, which may augment engagement
dynamics.
2 Mosquito Alert: CS for public health
Mosquito Alert (MA) is a CS research program initially devised to monitor the
expansion in Spain of the Asian tiger mosquito (Aedes albopictus), a disease
carrying mosquito. Since the expansion of the Zika virus threat in 2016, MA in-
cluded the monitoring of the yellow fever mosquito (Aedes aegypti ). Both species
are world wide distributed, living in urban environments, and being specially
invasive and agressive vectors of tropical infectious diseases such as Dengue,
Chikungunya, Yellow Fever, and Zika.
Aside from its scientific goals (e.g. unveiling the population expansion mech-
anisms, forecasting vector and disease threats), a particular challenge for MA
4
Fig. 1. Mosquito Alert (left) home screen of the app showing the rank and category;
(right) web map showing the validated reports.
arises from its impact on the public health domain. MA is aimed to provide
reliable early warning information (in recently colonised areas), and real-time
management information (in areas where it resides) to public health adminis-
trators. Public health administrations at different organizational levels in the
territory, use MA to improve their surveillance and control programs with the
goal of decreasing mosquito populations, specially in urban areas. Because of all
this, MA is designed as a multi-platform system structured as follows:
1. The MA smartphone app (freely available for Android and iOS), by means
of which citizen can send reports of observations of mosquitoes (and their
breeding sites) potentially belonging to disease vector species (namely the
Asian Tiger and the Yellow Fever mosquito).
2. The corresponding server-side functionality (Django, SQL) managing the
reception and storage of data, along with an ever evolving set of tools for the
management and analysis of the data, including machine-learning algorithms
to help automating the validation of information [13].
3. A private platform called Ento Lab. This is a restricted access service through
which a set of experts can make a previous filtering of inappropriate reports
and classify the rest as either positive or negative ones. Only classified reports
are afterwards made visible to the rest of the services.
4. Another private platform named Managers Portal which grants on-demand-
access to stakeholders (e.g. public health administrations, mosquito control
services, private mosquito control companies), open GIS tools to visualize all
the available data in the portal (including their own imported management
data), and the possibility to directly comunicate control actions through the
app.
5. A public platform http://www.mosquitoalert.com, providing data and visu-
alization tools to all the public via interactive maps, where participants can
find their individual contributions validated by the experts (Figure 1, right).
5
The direct involvement of public health institutions make citizen truly con-
scious of the usefulness of their contributions (much beyond science). Triggering
mosquito control actions in the territory through MA participation constitutes
the necessary reward to keep citizens engaged in the research program.
Table 1. Summary of Mosquit Alert’s data-base (2015-2016).
total NC hd -2 -1 0 +1 +2
adult 4177 19 188 429 128 655 1317 1441
% 0.00 0.05 0.10 0.03 0.16 0.32 0.34
total NC hd -1 0 +1
bSite 1172 564 160 90 129 229
% 0.48 0.14 0.08 0.11 0.20
The work described in this paper is based on data corresponding to the last
two years (2015-2016) of MA, summarized in Table 1, with more than 30000 app
downloads, 2993 active users and 5349 reports submited. Reports are of type
adult (4177) correponding to observations of adult mosquitoes (either Asian
tiger or Yellow Fever) or bSite (1172) corresponding to potential mosquitoes’
breeding sites. Reports of type adult and bSite are reviewed by experts who
manually label them. Reports of type adult are labelled as: −2, definitely not
corresponding to the species of interest; −1, probably not corresponding to the
species of interest; 0, can not tell; 1, probably corresponding to the species of
interest; and 2, definitely corresponding to the species of interest. Reports of
type bSite are labelled as: −1, does not correspond to a breeding site; 0, can not
tell; and +1, does correspond to a breeding site. NC stands for not-classified
reports which either do not provide an image or have not yet been reviewed by
the experts. The hd stands for hidden reports which correspond to reports with
improper images that are hidden by the experts (not shown in the map).
3 Methods
Let’s consider a Bayesian network describing the characteristic flow in a CS re-
search program, what we call the User-Action-Validation (UAV) network (Fig-
ure 2, left, lightgrey nodes). In our particular case, the nodes of the UAV network
represent the following:
– Users are participants of the CS research program aggregated in a variable
U = {1, . . . , k} specifying an ordinal set of user-classes with increasing levels
of expertise (e.g. beginner, advanced, expert). Beyond the intuition that U
should be a discrete ordinal variable, we have no prior idea of the optimal
number of user classes we should define nor about their prior distribution.
– Actions consist in submitting reports either of type adult or bSite. Thus, we
define A = {adult, bSite} such that P (A|U ) specifies the probability that a
6
Fig. 2. The User-Action-Validation Bayesian network to model users’ expertise on the
Mosquito Alert’s platform.
user of a given class submits a report of a particular type. We assume that
each submitted report is an independent event.
– Validations are expressed in terms of ratings and each kind of action can
be rated with a different number of discrete rating levels. Thus, we de-
fine a variable V such that P (V |A) specifies the probability that an action
of a given type gets the corresponding rating (or, transitorily, no rating),
namely (V |A = adult) = {N one, hd, −2, −1, 0, 1, 2} and (V |A = bSite) =
{N one, hd, −1, 0, 1}.
From the joint distribution P (U, A, V ) and by simple Bayes’ rule we have,
1
P (U |A, V ) = P (V |A) P (A|U ) P (U )
P (A, V )
where P (A, V ) is just a normalization constant. So, if we have some means to
estimate a prior distribution P (U ), and the conditional distributions P (A|U )
and P (V |A), we can evaluate the posterior distribution of user-class expertise
for a user y given the observed sequence of actions/validations S (y) (Figure 2,
right),
Y
P (U |S (y)) = P (U |A, V )
A,V ∈S(y)
3.1 Guessing a prior P (U )
We start by guessing a number of user expertise levels. For each user we consider
three features regarding to the sequence S (y): (i) the quantity of reports, (ii) the
quality of the reports, and (iii) a user’s mobility index mI describing the average
area covered by the user defined as the variance of the pairwise geolocation
distances between the reports,
1 X 2 2
mI (y) = [ (px − qx ) + (py − qy ) ] (4)
2 |S (y) |2
(p,q)∈S(y)
7
where (px , py ) , (qx , qy ) are the geolocation coordinates.
Based on these features we define the following proxy variables of the user-
class U (Figure 2, left, darkgrey nodes): (i) a quantitative proxy aggregating
users sending less or more than a given number θ1 of reports (N = {less, more});
(ii) a mobility proxy aggregating users with a mobility index lower/higher than
a given value θ2 (M = {lower, higher}); (iii) a quality proxy aggregating re-
ports in four categories: hidden, low quality (those labeled as (−2, −1)), medium
quality (those labeled as 0), and high quality (those labeled as (1, 2)), (Q =
{hidden, low, medium, high}, we do not count here not-classified reports). Note
that (i) and (ii) account for the attitude of the participants, while (iii) accounts
for their skills, and both aspects are deemed important. The joint combination
of the above three proxys results in a primary partition of the users’ expertise
space into 16 categories summarized in Table 2. The threshold values where se-
lected by looking at the corresponding histograms and taking the values that
yield the most possible balanced distribution.
Table 2. Join distribution of the proxy variables, P (N, M, Q), resulting in 16 expertise
categories. Threshold values: θ1 = 2, θ2 = 10−9 .
N ≤ θ1 , M ≤ θ2 N ≤ θ1 , M > θ2 N > θ1 , M ≤ θ2 N > θ1 , M > θ2
hiddden 0.0427 0.0220 0.0004 0.0086
low 0.0974 0.0132 0.0027 0.0228
medium 0.0853 0.0211 0.0065 0.0519
high 0.2685 0.0707 0.0322 0.2541
By looking at this table, we should now infer a set of user classes with increas-
ing levels of expertise. We prioritize as following: (i) the quality of the reports
before the quantity (low quality reports just result in a waste of experts’ time);
(ii) the quantity of reports before the mobility index of the users (we give the
lowest priority to the mobility index because the meaning of this variable is
double folded: for surveillance purposes it is important that participants send
reports covering the broadest geographical area possible, but for control pur-
poses it is also important that participants keep sending reports within their
neighbourhoods). Also, we are not looking here for a fine grain discretization of
the expertise space. Taking into account the unbalances present in Table 2 it
looks reasonable to impose an ordering of the 16 expertise categories into a set
of k = 6 user-classes, i.e U = {1, . . . , 6}, as shown in Table 3. Tables 2 and 3
together express a joint distribution P (N, Q, M, U ) from which the prior P (U )
follows straightforwardly by marginalization,
X
P (U ) = P (N, Q, M, U ) (5)
N,Q,M
and we get a temptative prior for the user-class variable (Table 4).
Having defined the user classes we know the user-class value of each report,
and we can make estimations (maximum a posteriori, MAP) for the action con-
8
Table 3. Definition of user classes
N ≤ θ1 , M ≤ θ2 N ≤ θ1 , M > θ2 N > θ1 , M ≤ θ2 N > θ1 , M > θ2
hiddden 1 1 1 2
low 2 2 3 3
medium 3 4 4 4
high 5 6 6 6
Table 4. Expertise-class prior distribution, P (U )
U 1 2 3 4 5 6
0.0648 0.1190 0.1106 0.0792 0.2691 0.3573
ditional distribution P (A|U ) (Table 5) and also for the validation conditional
distributions P (V |A = adult) and P (V |A = bSite) (Table 6).
3.2 Computing the posterior P (U |A, V )
Applying Bayes’ rule we compute the posterior distribution P (U |A, V ),
1
P (U |A, V ) = P (V |A) P (A|U ) P (U ) (6)
W
Pk
where W = i=1 p (ui |A, V ). Equation 6 evaluates the probability that an action
(report) of a given type, with a given validation (rating), belongs to a particular
user-class (Table 7).
Note that, in our Bayesian approach ratings become a fuzzy qualification
of the user-class. Also note that this is indeed a two parameter model (θ1 , θ2 )
allowing a degree of control over the user-class prior and, ultimately, over the
user-class posterior distributions, (i.e. we can push the classification of reports
to a lower/upper user-class by increasing/decreasing either one or both of the
parameters (this is shown later in Figure 4).
Table 5. Action conditional distribution, P (A|U )
U 1 2 3 4 5 6
adult .3454 0.7994 0.8456 0.4234 0.9339 0.9144
bSite 0.6546 0.2006 0.1544 0.5766 0.0661 0.0856
1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
3.3 Computing users’ scores.
Note that, so far, our UAV-network (Figure 2 , left) just yields the user-class
distribution of a single action, not the users’ scoring that we aim. To compute
9
Table 6. Validation conditional distribution, P (V |A, U )
adult reports
U 1 2 3 4 5 6
NC 0.0059 0.0059 0.0059 0.0059 0.0059 0.0059
hd 0.9643 0.0559 0.0022 0.0033 0.0008 0.0006
-2 0.0060 0.7326 0.1401 0.0033 0.0008 0.0006
-1 0.0060 0.1996 0.0657 0.0033 0.0008 0.0006
0 0.0060 0.0020 0.7817 0.9778 0.0008 0.0006
+1 0.0060 0.0020 0.0022 0.0033 0.5511 0.4135
+2 0.0060 0.0020 0.0022 0.0033 0.4397 0.5780
1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
bSite reports
U 1 2 3 4 5 6
NC 0.4742 0.4742 0.4742 0.4742 0.4742 0.4742
hd 0.5153 0.0932 0.0063 0.0064 0.0060 0.0035
-1 0.0035 0.4193 0.1837 0.0064 0.0060 0.0035
0 0.0035 0.0067 0.3294 0.5066 0.0060 0.0035
+1 0.0035 0.0067 0.0063 0.0064 0.5079 0.5152
1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
Table 7. User-class posterior distributions
adult reports
L NC hd -2 -1 0 +1 +2
U =1 0.0272 0.7810 0.0016 0.0052 0.0012 0.0005 0.0004
U =2 0.1157 0.1926 0.8351 0.7363 0.0018 0.0007 0.0006
U =3 0.1137 0.0074 0.1571 0.2382 0.6856 0.0007 0.0007
U =4 0.0408 0.0040 0.0013 0.0042 0.3074 0.0004 0.0004
U =5 0.3055 0.0075 0.0025 0.0080 0.0019 0.5050 0.3683
U =6 0.3972 0.0075 0.0025 0.0080 0.0019 0.4927 0.6296
1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
bSite reports
L NC hd -1 0 +1
U =1 0.2389 0.8849 0.0107 0.0050 0.0058
U =2 0.1346 0.0901 0.7252 0.0054 0.0062
U =3 0.0963 0.0044 0.2273 0.1922 0.0042
U =4 0.2575 0.0119 0.0212 0.7900 0.0115
U =5 0.1003 0.0043 0.0077 0.0036 0.3545
U =6 0.1724 0.0044 0.0078 0.0037 0.6178
1.0000 1.0000 1.0000 1.0000 1.0000
the scores we consider the report sequences S (y) (Figure 2, right). To add some
dynamics to the model we consider a third parameter θ3 (a cutoff factor ) that
limits the sequences to the last θ3 reports. Assuming an iid sequence of reports,
the corresponding user-class posterior distribution is given by,
10
θ3
P0 Y
P (U |S (y)) = P (U |Aj , Vj ) (7)
Pw j=1
where P0 = (1/k, . . . , 1/k) sets a starting uniform user-class distribution and
Pk
Pw = i=1 P (ui |S) is a normalization factor. Afterwards, an expertise score
can be computed as the user’s expected user-class,
k
1 1X
X (y) = E[U ]P (U |S) = ui p (ui |S) (8)
k k i=1
Equation 8 yields a normalized score with a lower bound given by k1 E[U ]P0
which avoids a discouraging zero-score for new comers. Usually, the computa-
tion of Equation 7 is subject to numerical precision problems and therefore we
implement a log computation as,
θ3
X
log P (U |S (y)) = log P0 + log P (U |Aj , Vj ) − log Pw (9)
j=1
For gamification purposes, users are ranked based on their scores. Ties are
solved by mobility index. The rank position, not the score, is notified to the
users via the smartphone app, together with a quantile based category label as
either gold, silver or bronze (Figure 1, left).
In summary, we use the global evidence in the data base to guess the joint
distribution P (U, Q, N, M ) and estimate a prior P (U ), from which we can de-
rive the posteriors P (U |A, V ). Afterwards, we use the evidence observed for
each particular user S (y), to evaluate the posterior distribution P (U |S (y)) and
compute a score for that user. Essentially, our scoring model is a naı̈ve Bayes
classifier where the number of features varies with the number of reports used
to qualify the user. The larger the number of reports, the better the profiling
of the user. Because we use the global evidence to estimate the user-class dis-
tribution, the scores change dynamically as the contents of the data base grow
and all individual expertise scores are in part dependent on the overall average
performance.
4 Results
Based on the user-class posterior distributions (Table 7) and applying Equa-
tions 9 and 8 we get the results shown in Figure 3. In the x-axis, users are
ranked by score (blue line). Scores are plotted together with scaled versions of
the mobility index (yellow), the number of breedingSite reports (cyan) and the
number of adultMosquito reports (magenta). The scoring yields several plateaus
corresponding to typical number of submitted reports. We highlight (darkgrey
rows in Table 8) the position of users who only submitted one report, classified
as +1 (positions 1494:2053, 560 users), or classified as +2 (positions 2244:2684,
11
Fig. 3. Ranked score plot
Table 8. Ranked score table
adult bSite
rank score hd -2 -1 0 +1 +2 hd -1 0 +1
1 .16668463 4
2 .16684409 3
3 .16728058 4
26:112 .19559913 1
116:201 .21448389 1
221:245 .33933867 2
263:553 .36258186 1
555:628 .38297512 1
629:670 .38557248 1
730:756 .52790362 2
764:1064 .55214793 1
1069:1278 .58333333
1284:1321 .63213868 1
1356:1390 .82074814 1 1
1494:2053 .91446382 1
2139:2205 .92599170 1
2244:2684 .93735892 1
2823:2882 .95749959 2
2892:2899 .95809921 2
2911:2926 .97219402 3
2936:2937 .97351476 3
2944:2952 .98252239 4
2962:2965 .98930969 5
2978 .99617983 7
2984 .99888390 9
2991 .99999356 1 19
2992 .99999384 18 1
2993 .99999632 20
12
Fig. 4. Effect of the 3 threshold parameters: (left) effect of the number of reports, θ1 ,
and the mobility index, θ2 ; (right) effect of the cutoff factor, θ3
441 users), which correspond to the largest plateaus in Figure 3. As expected,
the larger the number of positive reports the higher the rank (Table 8 bottom
rows), and the larger the number of negative reports the lower the rank (Table 8
top rows), with hidden reports being strongly punished.
We also analyze the effect of the threshold parameters (θ1 , θ2 , θ3 ) (Figure 4).
In the x-axis, participants are ranked by score (blue line). The red line depicts the
score coresponding to a different value of the threshold parameters. In general,
the scores do not change much in terms of value though sudden breaks in the
increasing trend of the red line reveal users whose position in the ranking has
been affected by the change of the parameter value. The plateaus remain almost
invariant and we only appreciate some changes of position at the borders of the
plateaus. Increasing (decreasing) θ1 and θ2 (Figure 4, left) together, move reports
to the left (right) columns of Table 2 and consequently force a change in the prior
distribution. As a result, the plateaus are globally pushed lower (higher). This
change is propagated to the posteriors and originates also the rank changes that
can be observed at the borders of the plateaus (Figure 4, left). In the case of
θ3 we observe that by not looking so far in the past (Figure 4, right), some
low rank users are upgraded (users who clearly improved their performance over
time) while some high rank users are downgraded (users who worsened their
performance over time).
The dynamics of the scoring are also shown. In Figure 5 (top) we simulate
the evolution of the score (blue line) and rank (red line) for a particular user
in a static situation where nothing is changing, no new users are coming and
no reports are submitted by third users. Each submitted report is shown as a
coloured dot, where the color indicates the validation value of the report. It is
apparent that good/bad reports push the score/rank up/down. Figure 5 (bot-
tom) shows the evolution of the score in a realistic situation to show the effect of
third users’ actions or new comers to MA. Note the double effect of the overall
dynamics, inducing soft fluctuations in the score but really significant changes
in the rank. The stronger dynamics of the rank makes it much more effective for
gamification purposes.
13
Fig. 5. Score dynamics: (top) simulating a static situation in which the rest of partici-
pants do not perform any action; (bottom) real situation with new reports submitted by
third participants and new participants joining the Mosquito Alert research program.
Fig. 6. Comparison of scores: (left) UAV-network vs. Bayesian Rating; (right) UAV-
network vs. Dirichlet reputation system with different values of C.
14
We also show a comparison of our scores with Bayesian Rating and Dirich-
let reputation scores (Figure 6). The scores used in the comparison have been
computed taking into account only the reports of type adult. In this way we
avoid to analyze second order effects due to weight-averaging of adult and bSite
reports, given that in BR or DR these must be independently computed and
combined later on. In the x-axis, participants are ranked by our score (blue
line). BR scores (Figure 6, left, magenta line) are clearly affected by the weight
of the overall average rating (note the scale of the right y-axis). However, BR
still yields the same plateaus and we only observe slight ranking changes at the
borders of the plateaus. These changes are due to the differences in the leverage
of the rate values (i.e. the values P (U |A, V ) in Tables 7 versus the rating levels
r = {1, . . . , k} used in Equation 1). DR scores (Figure 6, right) are computed
for different values of the C constant. It is clear that C is playing the role of the
overall average factor in BR, but DR gives us some control over it. The most im-
portant plateaus are also found, but the differences at the borders of the plateaus
are more significant. Notably, there is a great difference in the sensibility of our
model in comparison to BR and DR. In this context, sensibility represents a
better responsiveness of the scoring in relation to participants actions, which we
consider it to be a good property to improve participants’ engagement in CS
research programs.
5 Discussion
The model we propose is similar to a naı̈ve-bayes classifier where the number of
feature nodes varies with the number of actions performed (i.e. reports submit-
ted) by the user. Starting from a uniform user-class distribution, each validated
report contributes with new evidence to refine the profiling of the user.
The key issue of our approach is to estimate a user-class prior that suffices
for our scoring purpose. We suggest to select a set of proxy variables of the user-
class, with a clear semantics in terms of user expertise, to make a guess about
this prior. Nonetheless, any alternative to compute the prior can be considered
and applied as well. Anyway, it is crucial to make a guess of the prior that leads
to a well balanced (as much as possible) prior and to well behaved (as much
as possible) posteriors (i.e. good ratings favouring higher user classes and bad
ratings favouring lower user classes). If the probability mass distribution of the
posteriors is not in clear correlation with the user classes the behaviour of the
algorithm can become non-monotonic with respect to increasing evidence about
a certain class. Thus, this step must be carefully considered.
In the same way that it is not good to score new users excesively low, it is also
not good to score them excessively high. The reason to initialize the score with a
unifom instead of the prior user-class distribution is that the later will usually be
unbalanced, in our case, clearly unbalanced towards the high expertise classes
(Table 4), and consequently users with no validated ativity would be ranked
either excesively high or excessively low. Using the prior to initialize the score,
not-active users get a score of 0.74011666 (i.e. P (U |S) = P (U ) in Equation 8),
15
while using a uniform distribution their score is 0.58333333 (arround 0.5) and
they are positioned by the middle-low part of the rank (rank positions 1069:1278
in Table 8) which is fairly reasonable.
As scores are relative to the performance of the whole community, scores
are quite dynamic. As participants increase their expertise, all good scores are
globally pushed higher. Nevertheless, it is the rank what is ultimately notified
to the participants, thus along a period of no activity, a participant might be
downgraded with time. Furthermore, in periods of no activity the score can in-
deed increase if better positioned participants suddenly start sending reports of
low quality. These unexpected dynamics could easily generate some confusion or
disappointment among the participants. We avoid this situation by giving the
basic hints of our scoring system in the project’s web page 5 where, indeed, we
promote the gamification side of these features in order to use them in our favour.
Alternatively, unexpected dynamics as described above could be controlled by
implementing an age weighted rating as proposed in [9]. In our case, this solution
should be implemented with special care because of the seasonality of mosquito
population and, consequently, minimal report activity during winter and spring.
This long periods of minimal activity would uniform all scores and many expe-
rienced participants might feel disappointed. At the moment, our decision is to
keep participants’ scores from one season to another.
With respect to BR and DR, while essentially capturing the same concept
of rating-based reputation, our model shows a much higher sensibility to the
observed evidence, and a good balance of both, evidence of quality (the rates
themselves) and quantity of evidence (the number of ratings). The reason lies in
the way that evidence is cumulated, i.e. by multiplication (Equation 8) instead of
by addition as in BR (Equation 1) and in DR (Equation 2). A larger sensibility
results in a stronger responsiveness to specific participant actions. Augment-
ing engagement dynamics with more sensible reputation systems may probably
bind better the participants to the long term goals of CS research programs.
Furthermore, our model decouples action validation from participant scoring by
means of an integrative and unified treatment of any action under consideration,
independently of the rating system used for each type of action.
By summer 2017, MA is going to collect extra data from participants with
a recently added tool designed to reinforce citizen participation in the research
program, whilst easing the experts’ validation task. This new tool, natively in-
corporated to the app, allows citizens to validate mosquito and breeding site
images sent by third users and challenge their expertise in identifying mosquito
species. This new action will provide valuable information in terms of partici-
pants’ expertise, based on a binary rating system, i.e. right/wrong. Given the
structure of our scoring model, this information can be readily translated into a
new user-class posterior distribution and incorporated to the scoring algorithm.
5
http://www.mosquitoalert.com/en/project/send-data/
16
6 Conclusion
We propose a novel reputation system based on a Bayesian network representing
the characteristic flow typically present in CS research programs where partic-
ipants are expected to perform actions that are validated later on (i.e. user,
action, validation), what we call the UAV network. In this network, the users
node represents an aggregation of participants into expertise classes. The key
issue of our approach is to estimate a prior for the user-class distribution that
suffices for our scoring pourpose. We suggest to select a set of proxy variables of
the user-class, with a clear semantics in terms of user expertise, to make a guess
about this prior. However, any other means to get a valid estimate of the prior
can readily be used. With respect to Bayesian rating and the Dirichlet reputa-
tion models, our approach presents some advantages: (i) is more responsive to
the observed evidence, and thus, it bridges better participants with their actions,
(ii) it decouples action rating from user scoring, providing a unified processing
of any action under consideration, no matter the number of rating levels defined
for each, and (iii) it yields a better balance of both, evidence of quality (the
rates themselves) and quantity of evidence (the number of ratings). As a proof
of concept this model is implemented as part of the Mosquito Alert CS research
program.
7 Acknowledgments
We would like to thank the Mosquito Alert team for continuous effort and sup-
port and the Mosquito Alert community for its unvaluable cooperation. This
work is part of Mosquito Alert CS program research funded by the Spanish
Ministry of Economy and Competitiveness (MINECO, Plan Estatal I+D+I
CGL2013-43139-R) and la Caixa Banking Foundation. Mosquito Alert is cur-
rently promoted by la Caixa Banking Foundation.
References
1. N. Eyal and R. Hoover. Hooked: How to Build Habit-Forming Products. Portfolio
Penguin, 2014.
2. Randy Farmer and Bryce Glass. Building Web Reputation Systems. Yahoo! Press,
USA, 1st edition, 2010.
3. BJ Fogg. A behavior model for persuasive design. In Proceedings of the 4th In-
ternational Conference on Persuasive Technology, Persuasive ’09, pages 40:1–40:7,
New York, NY, USA, 2009. ACM.
4. Eric Friedman, Paul Resnick, and Rahul Sami. Manipulation-resistant reputation
systems. In Noam Nisan, Tim Roughgarden, Eva Tardos, and Vijay Vazirani,
editors, Algorithmic Game Theory, pages 677–698. Cambridge University Press,
2007.
5. Roy H.E., Preston M.J.O, C.D., Roy D.B., Savage J., Tweddle J.C., and Robinson
L.D. Understanding Citizen Science & Environtmental Monitoring. Final Report
on behalf of UK-EOF. NERC Centre for Ecology & Hidrology and Natural History
Museum, November 2012.
17
6. Ferry Hendrikx, Kris Bubendorfer, and Ryan Chard. Reputation systems. J.
Parallel Distrib. Comput., 75(C):184–197, jan 2015.
7. A. Irwin. Citizen Science: A Study of People, Expertise and Sustainable Develop-
ment. Routledge, 1995.
8. Audun Jøsang. Trust and reputation systems. In Alessandro Aldini and
Roberto Gorrieri, editors, Foundations of Security Analysis and Design IV: FOSAD
2006/2007 Tutorial Lectures, pages 209–245, Berlin, Heidelberg, 2007. Springer-
Verlag.
9. Audun Josang and Jochen Haller. Dirichlet reputation systems. In Availability,
Reliability and Security, 2007. ARES 2007. The Second International Conference
on, pages 112–119. IEEE, 2007.
10. Audun Jøsang and Roslan Ismail. The beta reputation system. BLED 2002 Pro-
ceedings, page 41, 2002.
11. L. Mui, M. Mohtashemi, and A. Halberstadt. A computational model of trust and
reputation for e-businesses. In Proceedings of the 35th Annual Hawaii International
Conference on System Sciences (HICSS’02), volume 7 of HICSS ’02, pages 188–,
Washington, DC, USA, 2002. IEEE Computer Society.
12. Colin Robertson. Whitepaper on citizen science for environmental research. 2015.
13. Antonio Rodriguez, Frederic Bartumeus, and Ricard Gavaldà. Machine learn-
ing assists the classification of reports by citizens on disease-carrying mosquitoes.
In Ricard Gavaldà, Indre Zliobaite, and João Gama, editors, Proceedings of the
First Workshop on Data Science for Social Good co-located with European Con-
ference on Machine Learning and Principles and Practice of Knowledge Dicovery
in Databases, SoGood@ECML-PKDD 2016, Riva del Garda, Italy, September 19,
2016, volume 1831 of CEUR Workshop Proceedings. CEUR-WS.org, 2017.
14. Jordi Sabater and Carles Sierra. Review on computational trust and reputation
models. Artif. Intell. Rev., 24(1):33–60, September 2005.
15. J. Silvertown, M. Harvey, R. Greenwood, M. Dodd, J. Rosewell, T. Rebelo, J. An-
sine, and K. McConway. Crowdsourcing the identification of organisms: A case-
study of ispot. ZooKeys, 480:125–146, 2015.
18