=Paper=
{{Paper
|id=Vol-1229/dynak2014_paper8
|storemode=property
|title=Collaborative Filtering based on Dynamic Community Detection
|pdfUrl=https://ceur-ws.org/Vol-1229/dynak2014_paper8.pdf
|volume=Vol-1229
|dblpUrl=https://dblp.org/rec/conf/pkdd/AbdrabbahAA14
}}
==Collaborative Filtering based on Dynamic Community Detection==
https://ceur-ws.org/Vol-1229/dynak2014_paper8.pdf
Collaborative Filtering based on Dynamic
Community Detection
Sabrine Ben Abdrabbah, Raouia Ayachi, and Nahla Ben Amor
LARODEC, Université de Tunis, ISG Tunis,
2000, Bardo, Tunisia
abidrabbah.sabrine@gmail.com,raouia.ayachi@gmail.com,nahla.benamor@gmx.fr
Abstract. With the increase of time-stamped data, the task of recom-
mender systems becomes not only to fulfill users interests but also to
model the dynamic behavior of their tastes. This paper proposes a novel
architecture, called Dynamic Community-based Collaborative filtering
(D2CF), that combines both recommendation and dynamic community
detection techniques in order to exploit the temporal aspect of the commu-
nity structure in real-world networks and to enhance the existing community-
based recommendation. The efficiency of the proposed D2CF is dealt with
a comparative study with a recommendation system based on static com-
munity detection and item-based collaborative filtering. Experimental re-
sults show a considerable improvement of D2CF recommendation accu-
racy, whilst it addresses both of scalability and sparsity problems.
Keywords: Recommendation systems, Collaborative filtering, Dynamic
community detection, Time varying graphs
1 Introduction
Several types of recommenders have been proposed and can be categorized into
three major categories [16], namely content-based filtering, collaborative filter-
ing, and hybrid approaches.
In this work, we focus on collaborative filtering which predicts users inter-
ests/preferences from those of remaining users sharing similar tastes. Collabora-
tive filtering is considered as one of the most used techniques due to its efficiency
and its high accuracy . Typically, the recommendation process for this technique
starts when users express their preferences by rating items. The system analyzes
these ratings to determine the exact preferences of the user, then, matches the
active user’s preferences and the preferences collection to discover the category of
users having similar taste with the active user. Finally, the system recommends
a set of items for the active user according to the preferences of their similar
users.
From another side, the community detection presents a growing interest for
many researchers, especially in web applications. A panoply of community detec-
tion algorithms exist in literature and most of them focus on static community
detection, but recently the dynamic aspect of networks has sparked a new line of
research. Static community detection algorithms have been explored in collab-
orative filtering, based on the idea that community structure will enhance the
performance of recommendations [1, 3–6]. Nevertheless, these ones are not able
to deal with the dynamic aspect of real-world networks.
The purpose of this paper is to model the dynamic behavior of users in
recommender systems. We assume that users behaviors are learned from users’
ratings data and more specifically by looking firstly at the items which have been
rated by each user and then finding how to link items to each other over time in
order to build the dynamic network of items. Our major contribution consists in
presenting a novel approach named Dynamic Community-based Collaborative
Filtering (denoted D2CF for short) capturing dynamic communities of items
which present the evolution of users interests and preferences over time to over
recommendations more suitable for real-world networks.
This paper is organized as follows: Section 2 presents the basic concepts of
recommendation and community detection. Related work is provided in Sec-
tion 3. Our proposed architecture D2CF is presented in Section 4. Section 5 is
dedicated to the experimental study.
2 Basic concepts for recommendation and community
detection
This section gives a brief overview on both recommendation systems and com-
munity detection.
2.1 Recommendation
Recommender systems have been proposed to address the information overload
problem by filtering the relevant data and suggesting items of potential interest
to users. Formally, in a typical recommendation system, there is a set of users
U and a set of items I. The task of recommender system is to predict user’s
preferences P for each item in I. The output is a list L containing items with
the highest preference values. Content-based filtering, collaborative filtering and
hybrid approaches are three major categories of recommendation methods:
– Content-based approach selects items based on their content along user’s
profile. Its principle is to recommend items similar to the ones that the user
has preferred in the past.
– Collaborative filtering approach infers user’s preferences from remaining users
having similar tastes. Methods pertaining to this approach can be divided
into user-based and item-based methods.
– Hybrid approach combines content-based and collaborative filtering methods,
in other words it takes into account both the users and the items properties.
In this work, we will focus on the most widely used recommendation method,
namely Item-based Collaborative Filtering [19]. In such a case, the user’s pref-
erence is predicted on an item using the average ratings of similar items by the
same user as expressed in Equation (1) where S represents the most similar
items to the item i, s(i, j) denotes the similarity degree between items i and j
and ru,j corresponds to the rating of user u on item j.
P
j2S s(i, j) ru,j
Pu,j = P (1)
j2S |s(i, j)|
The similarity value can be calculated in many ways. Common methods are
Cosine similarity and Pearson Correlation [19].
Due to the nature of the data used in collaborative filtering, this approach
su↵ers, as the case of all methods, from one or more weaknesses such as the
cold start problem when a new user starts with an empty profile, the sparsity
problem occurring when available data are insufficient for identifying similar
users, and the scalability problem when there is an excessive information of users
and items. To overcome these problems, several recommendation methods have
been implemented using di↵erent techniques. We cite in particular, clustering
techniques [13], Bayesian techniques [12] and community detection techniques
[1–4].
2.2 Community detection
Community detection techniques aim to find subgroups where the amount of
interactions inside the group is more than the interaction outside it, and this
can help to understand the collective behavior of users.
The communities identification process depends on the nature of networks,
either static or dynamic. Static networks are basically constructed by aggregat-
ing all observed interactions over a period of time and representing it as a single
graph. Dynamic networks, also called time varying graphs can be either a set of
independent snapshots taken at di↵erent time steps [7, 8] or a temporal network
that represents sequences of structural modifications over time [9]. In what fol-
lows, we present both static and dynamic community detection.
Static community detection: A panoply of community detection algorithms
exists in literature. The first idea using static networks was proposed by Girvan
and Newman [15]. It is based on a modularity function representing a stopping
criterion, aiming to obtain the optimum partitioning of communities. In the
same context, Guillaume et al. [11] have proposed Louvain algorithm to detect
communities using the greedy optimization principle that attempts to optimize
the gain of modularity. Rosvall and Bergstrom [17] have presented Infomap,
considered as a solution to the simplest problem of static and non-overlapping
community detection. The mentioned algorithms are not able to detect overlap-
ping communities where a node can belong to more than one community in the
same time. To ensure this basic property, Palla et al. [14] have proposed the
Clique-Percolation Method (CPM) to extract communities based on finding all
possible k-cliques in the graph. This method requires the size of the cliques in
input.
Dynamic community detection: Several researchers explored the dynamic
aspect of networks to identify communities structure and their development
over time. Hopcroft et al. [7] have proposed the first work on dynamic com-
munity detection which consists in decomposing the dynamic network into a set
of snapshots where each snapshot corresponds to a single point of time. The
authors applied an agglomerative hierarchical method to detect communities in
each snapshot and then they matched these extracted ones in order to track their
evolution over time. Palla et al. [8] have used the (CPM) method of static com-
munity detection to extract communities from di↵erent snapshots. Then, they
tried to look for a matching link between them to detect their structural changes
over time. Methods applying static algorithms on snapshots cannot cover the
real evolution of communities structures over time because it seems harder for
these methods to recognize the same community from two di↵erent time steps
of network.
To overcome this problem, new studies have exploited another representation
of data that takes into account all temporal changes of the network in the same
graph. We cite, in particular, the intrinsic Longitudinal Community Detection
(iLCD) algorithm proposed by Cazabet et al. [9]. The algorithm uses a longi-
tudinal detection of communities in the whole network presented in form of a
succession of structural changes. Its basic idea was inspired from multi-agent sys-
tems. In the same context, Nguyen et al. [10] have proposed AFOCS algorithm to
detect overlapping communities in a dynamic network N composed of the input
network structure N0 and a set of network topology changes {N1 , N2 , ..., Nn }.
3 Related works
There has recently been much research on merging community detection and
recommender systems in order to provide more personalized recommendations
related to users belonging to the same community. In fact, community-based
recommendation is a two-step approach. The first step consists in identifying
groups in which users should share similar properties and the second step uses
the community into which the target user pertains to recommend new items.
Using static community detection algorithms, Kamahara et al. [2] have pro-
posed a community-based approach for recommender systems which can reveal
unexpected user’s interests based on a clustering model and an hybrid recommen-
dation approach. In the same context, Qin et al. [3] have applied CPM method
on the Youtube Recommendation Network of reviewers to detect communities of
videos. These latters are used to provide the target user by a local recommenda-
tions which consists in recommending videos pertaining to the same community
of the video watched by him. This approach aims to propose a more diverse list
of items for target user. Another aim behind incorporating community detec-
tion to recommendation is to provide a solution to the cold start problem, and
this idea was proposed by Sahebi et al. [4] while applying Principal Modularity
Maximisation method to extract communities from di↵erent dimensions of social
networks. Based on these latent communities of users, the recommender system
is able to propose relevant recommendations for new users.
Qiang et al. [1] have defined a new method of personalized recommenda-
tion based on multi-label Propagation algorithm for static community detection.
The idea consists in using the overlapping community structures to recommend
items using collaborative filtering. More recently, Zhao et al [5] proposed the
Community-based Matrix Factorization (CB-MF) method based on communi-
ties extracted using Latent Dirichlet Allocation method (LDA) on twitter social
networks. In [6], authors focused on community-based recommendation of both
individuals and groups. They used the Louvain community detection method on
the social network of movies building from the Internet Movie Database (IMDb)
in order to provide personalized recommendations based on the constructed com-
munities.
These methods only deal with static networks, derived from aggregating data
over all time, or taken at a particular time. The accumulation of an important
mass of data in the same time and in the same graph can lead to illegible graphs,
not able to deal with the dynamic aspect of real-world networks. To take into ac-
count the evolution of users behaviors over time using a kind of community-based
dynamic recommendation, a first attempt was established by Lin et al. [16]. The
main idea consists in providing a dynamic user modeling method to make rec-
ommendations by taking into consideration the dynamic users’ patterns and the
users’ communities. This approach is limited since it uses a manual method to
identify communities, which is not efficient especially when we deal with strongly
evolving and large networks. More recently, Abrouk et al. [20] proposed to use
the fuzzy k-means clustering from time to time to dynamically detect the users’
interests over time. Then, they exploited these formed communities to determine
user’s preference for new items with regard to the updated users’ ratings. In [21],
author proposed an article recommender system to recommend documents for
users based on the same members of communities which are identified accord-
ing to their interests while browsing the web. The detection of users’ interests
is repeated continuously in subsequent time intervals in order to deal with the
dynamic aspect of new portals. In both the previously presented methods, ap-
plying clustering techniques for time to time cannot cover the real evolution of
community structure over time. In fact, several structural changes may occur
and get lost without being detected. Besides, the temporal complexity of these
methods increases in large networks.
Aiming to benefit from the whole advantages of the dynamic community de-
tection process as part of recommender systems, we propose a global architecture
allowing to ensure this combination as detailed below.
4 Proposed architecture
The proposed architecture, called Dynamic Community-based Collaborative Fil-
tering, denoted by D2CF, is based on three main steps as shown in Figure 1.
0+%%1#&)&(2
!"#$%&' #()*+,-
3456,(7,+'(22 945!"#$%&' '+%%1#&)"
2)(7 :()(')&+# 2)(7 ;4/('+%%(#:$)&+#52)(7
!"#$%
&'(#)$* ./
+('(
Fig. 1. D2CF architecture
4.1 The pre-processing step
This step consists in building a dynamic network in which the evolution of the
users’ interests (i.e. the set of items be looked at or ranked) over time is repre-
sented. We assume that if a user does not rate a given item then this latter is not
yet watched by him. It is important to note that, in this work, the nodes are the
items and not the users and hence the communities are groups of items. The in-
teractions between nodes are modeled using the co-ratings relationship. Indeed,
two nodes interact with each other if at least one user gives the same rating
to both of them in the same time. With the aim to prepare the list of network
changes over time, we have adapted the method of temporal network building
[18] (i.e. An edge is established between two nodes if these ones have interacted
with each others at least N times over a period of P days). The values of N and
P control the set of edges adding and removal in the network and depend not
only on the network topology but also on the information that we want to ex-
tract in order to create semantic links between the network nodes (e.g. messages
interactions seem to be more intense than calls interactions). If over a period of
P days, there are fewer than N interactions between two nodes, the edge will
be automatically removed. These changes (edges removal and edges creations)
can be represented either in the same graph as temporal network or in di↵er-
ent graphs as a sequence of snapshots where each one corresponds to network
changes over a specific time period (i.e. one hour, one day, ect.). The resulting
dynamic network of items is considered as a generic model that represents the
evolution of users interests over time.
4.2 Dynamic community detection step
Once the dynamic network (temporal network or a sequence of snapshots) is
constructed in the pre-processing step, we are now able to use it as input to
dynamic community detection method. Based on the network interactions, a
community can be defined as a set of items that are extremely related to each
other physically (strongly connected) and semantically (a learning pattern of
items that tends to have the same interests of several users over a period of
time). The advantage here is that a community is not restricted to item-related
topics but it contains various topics (e.g. Horror, Comedy, Romance, ect.). The
novelty in this work is to apply a dynamic community detection algorithm that
takes into account the evolution of the network behavior in order to obtain a
more appropriate community structure. To ensure this step, we choose to use
the iLCD algorithm [9]. This choice is justified by the fact that iLCD detects
communities in both static and dynamic networks depending on disposal data.
Moreover, this algorithm takes into account the evolution of the network which
enables to identify the dynamical communities more accurately. Such community
detection can be more powerful because this analysis matches with the reality of
networks. The algorithm deals with both large evolving networks and the overlap
of communities. Extracted communities are atomic in the sense that there are
not other relevant communities inside it. Finally, all operations in iLCD are
made at a local level (i.e. a community can interact with only the nodes that are
linked to and having at least one node in common), which can allow to ensure
that the complexity will not grow exponentially with the size of the network.
The life cycle of a community may be described via three phases:
– A new community is born when a new clique is formed in the graph.
– A modification of an internal community can lead to merge two communities
having at least one node in common or to split the main community into
two new communities.
– A community dies when there is no disposal nodes in its structure.
4.3 Recommendation step
In this step, the learned patterns (communities) will be exploited to help the
recommender system to predict the users’ future interests based on certain cat-
egories given by these communities. Firstly, a target item should be identified
for each user. The target item in this case is the item in which the active user is
more interested (The item with the highest rating value). The items that belong
to the communities of target item and that are not rated by the active user are
selected as candidate items. The list of top k recommended items for the active
user contains the k candidate items that have the highest predicted preferences.
Our objective is to compute the preference of the user u on the candidate item
i based on the items that belong to the communities of i. By doing this the
recommendation is restricted on the communities to which the candidate item
pertains. To this end, we propose an adaptation of the traditional item-based
Collaborative filtering method (See Equation (1)). In fact, instead of comput-
ing user’s preferences taking into consideration all items present on the whole
network to discover the most similar items to the candidate item, we only rely
on the items that pertain to candidate item’s communities extracted from the
dynamic network as shown in Figure 2.
Target item
Item-based CF D2CF approach
Fig. 2. The principle of user’s preference computation taking into account the commu-
nity structure in the network
Thus, we can formally define the preference prediction as follows:
P
j2C s(i, j) ru,j
Pu,i = P (2)
j2C |s(i, j)|
where C is the set of items pertaining to the community of i, ru,j is the
rating given by the active user u to the item j and s(i, j) is the similarity degree
between items i and j.
We propose to compute the similarity s(i, j) using the Pearson correlation
similarity measure [19].
Finally, the recommendation list contains the candidate items i having the
highest preference values Pu,i .
In the case where the user is new (no ratings history), the target item of
this user is learned by browsing item in the recommender system. We select
then the candidate items that belong to the communities of the target item. The
recommendation list contains the candidate items which are ranked according
to their similarities relative to target item.
5 Experimental study
To evaluate the e↵ectiveness of D2CF, we propose to use the movieLens dataset
available through the movieLens website (http://movieLens.umn.edu). This dataset
contains in total 100.000 ratings collected by 943 users on 1682 movies, from 19-
09-1997 to 22-04-1998. The score of rating is ranged from 1 to 5. Each user has
rated at least 20 movies. The ratings information are timestamped. We suppose
that each movie is represented by a node and a community is defined as a set of
nodes. If a user rates a movie, this means that he is interested in watching it.
MoviLens data are represented as a sequence of temporal events in the following
way:
– user U1 rates movie I1 with 5 at T1 ,
– user U2 rates movie I1 with 3 at T1 ,
– user U2 rates movie I5 with 5 at T2 , etc.
To experimentally determine the impact of the training size on the quality of
the recommended movies, we propose to test three scenarios:
– Set 1 : For each user, we randomly select 90% of his ratings as instances in
the training set and the remaining ones will be used in the testing set.
– Set 2 : For each user, we randomly select 40% of his ratings as instances in
the training set and 10% will be used in the testing set.
– Set 3 : For each user, we randomly select 20% of his ratings as instances in
the training set and 10% will be used in the testing set.
The data selection should take into account the evolution over time, so that,
instances of the testing set should be chosen after those of the training set.
We run 10-fold cross-validation on data. The precision is defined to evaluate
the validity of a given recommendation list and it is formulated to detect the
average of the true recommendations relative to the total number of the proposed
recommendations. While the recall metric is defined as the ratio of the number
of recommended objects collected by users appearing in the test set to the total
number of the objects actually collected by these users.
The implementation of the whole system needs several parameters. Our
choices regarding the three steps of D2CF can be summarized as follows:
1. Pre-processing step: Our idea is to extract movies interactions in such a way
that we know what a movie has been assessed with another one with the
same score by the same producer in the same time. In fact, if an interaction
between two movies occurred more than N times over a period of P days,
an edge is established between them. We define the values of P and N such
a way that we conserve more links between nodes. After performing several
tests on the movieLens data, we set P and N respectively to 200 and 30 for
Set 1, 200 and 20 for Set 2 and 200 and 5 for Set 3. This means that, in
Set 1, an edge is established between two nodes, if these ones have interacted
at least 30 times over a period of 200 days and this edge is removed if less
than 30 interactions have occurred between them over a period of 200 days
after the edge creation date. The movies that are not very visible in the users’
ratings behaviors are considered as outliers. The outliers are the nodes that
are disconnected of the core of the network due to their low interactions with
other movies (i.e. there are less than N users who give the same rating for
both of movies).
2. Dynamic community detection step: In this stage, we are able to apply any
state of art dynamic community detection algorithm. In this experiment we
choose to use iLCD algorithm to extract communities from the temporal
network built above. Since the quality of resulting communities depends on
the threshold (i.e. is a parameter using as input in the community detection
step to determine the belonging or not of new added nodes in a community),
we choose the value 0.5 to obtain by the end of the process, overlapping, small
and dense communities.
3. Recommendation step: Using detected communities, we are now able to gen-
erate the top k recommendation list of movies to the active user. This step
requires both the user’s ID to look for his target item and the community
structure as input parameters to select the candidate items that may inter-
est the active user. Then, the predicted preference of each candidate item
is computed using Pearson correlation-based similarity measure. The items
which have the top k preference predictions are recommended to the active
user.
In order to evaluate both of the e↵ectiveness and efficiency of our proposed D2CF
approach we compare the performance of this one with the following methods:
– The Static Community-based Collaborative Filtering (denoted S2CF for
short), which is a static version of our proposed architecture. We keep the
same parameters used for the dynamic network without taking into account
the temporal dimension. This is possible since, as mentioned before, the
iLCD algorithm allows both static and dynamic community detection.
– The traditional item-based Collaborative Filtering with Pearson correlation-
based similarity measure available from Apache mahout library in Java.
The obtained results are summarized in Table 1. We can notice that our D2CF
method outperforms traditional recommendation methods: item-based collabo-
rative filtering and Collaborative filtering based on static community detection.
In fact in Set 3, our approach is able even with small set of users’ data to provide
users with a wealthy and varied recommendation list based on the communities
of users’ interests. The recall and precision values for both item-based and static
community-based collaborative filtering decrease as we increase the training set
size but our approach combining recommendation and dynamic community de-
tection still provide better recommendation quality as shown in Figure 5. In fact,
D2CF gives its best results (i.e. D2CF proposes for one user an average of 6 good
items out of every 10 recommended items while S2CD o↵ers an average of two
good items out of every 10 recommended items and finally item-based collabo-
rative filtering gives an average of 0,15 good items per user). These results show
that our approach is the best in the case of scalable data, which is explained by
the fact that the dynamic network learned by more users’ data better performs
the prediction of users’ preferences for unseen items.
Table 1. Precision and Recall values for Set 1, Set 2 and Set 3
Set 1 Set 2 Set 3
Approach Precision Recall Precision Recall Precision Recall
D2CF 0.603 0.687 0.3 0.49 0.223 0.34
S2CF 0.2 0.26 0.21 0.195 0.197 0.221
Item-based CF 0.015 0.02 0.084 0.091 0.18 0.2
D2CF approach presents a significantly improvement on recommendation
on both small and large sets. We can say that this approach addresses both
!"#$"%&'(" )*+()),+#"$)--"%,'&.)%/
(!
'!
&!
%! -+*./01)2345
6#45
$!
7#45
#!
"!
!
)*+," )*+,# )*+,$
Fig. 3. Impact of the dataset size on the recommendation quality
scalability and sparse problems and it is able to handle the real-world networks
by providing a dynamic recommendation based on dynamic communities.
6 Conclusion
In this paper, we propose a Dynamic Community-based Collaborative Filtering
approach that combines recommendation and dynamic community detection.
This approach is able to deal with real-world networks as it takes into account
the evolutionary aspect of the users’ interests over time. The experimental results
show that our proposed D2CF outperforms both of item-based collaborative
filtering and collaborative filtering based on static communities. As a future
work, we will explore the similarity computation process of users pertaining to
the same community in the recommendation context.
References
1. Qiang, H., Yan, G.: A method of personalized recommendation based on multi-label
propagation for overlapping community detection. In: the 3rd International Confer-
ence on System Science Engineering Design and Manufacturing Informatization, 1,
pp. 360–364. October (2012)
2. Kamahara, J., Asakawa, T., Shimojo, S., Miyahada, H.: A commynity-based recom-
mendation system to reveal unexpected interests. In: Multimedia Modelling Con-
ference, pp. 433–438. January (2005)
3. Qin, S., Menezes, R., Silaghi, M.: A recommender system for youtube based on its
network of reviewers. In: the IEEE International Conference on Social Computing,
(2010)
4. Sahebi, S., Cohen, W.: Community-based recommendations: a solution to the cold
start problem. In RSWEB’11, (2011)
5. G. Zhao, G., Lee, M. L., Hsu, W., Chen, W., Hu, H.: Community-based user rec-
ommendation in uni-directional social networks. IN: the 22th ACM international
conference on Conference on information knowledge management, pp. 189–198. Oc-
tober 2013.
6. Fatemi, M., Tokarchuk, L.: A community based social recommender system for
individuals groups. In: the 2013 International Conference on Social Computing (So-
cialCom’13), pp. 351–356. Sept (2013)
7. Hopcroft, J., Khan, O., Kulis, B., Selman, B.: Tracking evolving communities in
large linked networks. In: the national academy of sciences of the United States of
America, 1, pp. 5249–5253. (2004)
8. Palla, G., Barabasi, A., Vicsek, T.: Quantifying social group evolution. In: Nature,
446, pp. 664–667. April (2007)
9. Cazabet, R., Amblard, F.: Simulate to detect: a multi-agent system for community
detection. In: The 2011 ACM International Conference on Web Intelligence and
Intelligent Agent Technology(WI-IAT), 2, pp. 402–408. August 2011.
10. Nguyen, N., Dinh, T., Tokala, S., Thai, M.T.: Overlapping communities in dynamic
networks: Their detection and mobile applications. In: the 17th annual international
conference on Mobile computing and networking (MobiCom’11), pp. 85–96. Sept
(2011)
11. Blondel, V., Guillaume, J., Lambiotte, R., Lefebvre, E.: Fast unfolding of commu-
nities in larges networks. Journal of Statical Mechanics : Theory and Experiment.
(2008)
12. Condli↵, M. K., Lewis, D. D., Madigan, D.: Bayesian mixed-e↵ects models for
recommender systems. In: ACM SIGIR’99 Workshop on Recommender Systems:
Algorithms and Evaluation, (1999)
13. Sarwar, B. M., Karypis, G., Konstan, J., Riedl, J.: Recommender systems for large-
scale e-commerce: Scalable neighborhood formation using clustering. (2002)
14. Palla, G., Derényi, I., Farkas, I., Vicsek, T.: Uncovering the overlapping community
structure of complex networks in nature and society. In: Nature, 435(7043), pp. 814–
818. (2005)
15. Newman, M., Girvan, M.: Finding and evaluating community structure in net-
works. Phisical review E, 69(2). (2004)
16. Song, X., Lin, C., Tseng, B., Sun, M.: Modeling evolutionary behaviors for
community-based dynamic recommendation. In: the 2006 SIAM International Con-
ference on Data Mining, (2006)
17. Rosvall, M., Bergstrom, C.: Maps of random walks on complex networks reveal
community structure. In: the National Academy of sciences, 105(4), pp. 1118–1123.
(2008)
18. Cazabet, R.: Dynamic Community detection on temporal networks. PhD thesis,
Université Toulouse 3 Paul Sabatier. (2013)
19. Sarwar, B., Karypis, G., Konstan, J., Riedl, J.: Item-based collaborative filtering
recommendation algorithms. In: the 10th international conference on world wide
web, www’01, pp. 285–295. (2010)
20. Abrouk, L., Gross-Amblard, D., Cullot, N.: Community detection in the collabora-
tive web. International Journal of Managing Information Technology. pp. 1–9.(2010)
21. Hönsch, M.: Detecting user communities based on latent and dynamic interest on a
news portals. In: the 7th student research conference in informatics and information
technologies, 3, pp. 47-50. ACM(2011)