=Paper= {{Paper |id=Vol-1911/11 |storemode=property |title=Schema-aware Feature Selection in Linked Data-based Recommender Systems |pdfUrl=https://ceur-ws.org/Vol-1911/11.pdf |volume=Vol-1911 |authors=Corrado Magarelli,Azzurra Ragone,Paolo Tomeo,Tommaso Di Noia,Matteo Palmonari,Andrea Maurino,Eugenio Di Sciascio |dblpUrl=https://dblp.org/rec/conf/iir/MagarelliRTNPMS17 }} ==Schema-aware Feature Selection in Linked Data-based Recommender Systems== https://ceur-ws.org/Vol-1911/11.pdf

Schema-aware feature selection in Linked
Data-based recommender systems
(Extended Abstract)?

Corrado Magarelli1 , Azzurra Ragone1 , Paolo Tomeo2 , Tommaso Di Noia2 ,
Matteo Palmonari1 , Andrea Maurino1 , Eugenio Di Sciascio2
1
University of Milan Bicocca, P.zza Dell’Ateneo Nuovo, 1, 20126 Milano, Italy
{corrado.magarelli,azzurra.ragone,matteo.palmonari,andrea.maurino}@unimib.it
2
Polytechnic University of Bari, Via Orabona, 4, 70125 Bari, Italy
{paolo.tomeo,tommaso.dinoia,eugenio.disciascio}@poliba.it

Abstract. Semantics-aware recommendation engines have emerged as
a new family of systems able to exploit the semantics encoded in un-
structured and structured information sources to provide better results
in terms of accuracy, diversity and novelty as well as to foster the pro-
visioning of new services such as explanation. In the rising of these new
recommender systems, an important role has been played by Linked Data
(LD). However, as Linked Data is often very rich and contains many in-
formation that may result irrelevant and noisy, an initial step of feature
selection may be required in order to select the most meaningful portion
of the original dataset. Many approaches have been proposed in the lit-
erature for feature selection that exploit different statistical dimensions
of the original data. In this paper we investigate the role of the semantics
encoded in an ontological hierarchy via schema-summarization when ex-
ploited to select the most relevant properties for a recommendation task.

1 Introduction

In the last years we have witnessed a flowering of semantics-aware solutions for
Recommender Systems (RSs) exploiting information held in knowledge graphs,
as the ones available in the Linked Data (LD) Cloud. Several approaches using
LD to build RSs have been proposed in the literature. However, almost no one
tackles the issue of automatically selecting the best subset of LD-based features.
Usually, the feature-selection process is done manually by choosing the prop-
erties more ”suitable” for the scenario taken into account. For example, in a
scenario related to movies, properties as dbo:starring or dbo:director look
more relevant than dbo:releaseDate or dbo:distributor. As well as for the
music domain, properties as dbo:genre and dbo:writer look more important
than dbo:producer or dbo:recordedIn. However, without an automatic feature
selection process, the human intervention is required every time a new domain
?
An extended version of this paper has been published in [7]
is chosen, while it could be good to have a general way to select properties re-
gardless of the domain. In machine learning tasks there is the need to perform a
selection of features and this could not be straightforward when attributes are
embedded in a knowledge graph. In many graph-based recommendation systems
the knowledge exploration starts from the data and goes on following the rela-
tions between entities, without taking into account the knowledge lying in the
ontology and then in its class hierarchy. In this paper we investigate how ontolog-
ical schema summarization could be used as a feature selection technique for LD-
based recommender systems when features are represented by RDF properties
and compare the results with other ”classical” techniques for feature selection.

2 Feature selection and recommender systems

When dealing with recommender systems, a relevant task is to determine the im-
pact of a particular feature selection technique on the behavior of the underlying
algorithms. Indeed, some techniques can improve the accuracy of the recommen-
dation, some improves the diversity while others can provide a good trade-off
between diversity and accuracy. Among all the different feature selection tech-
niques available in the literature, in our experimental setting, we initially selected
Information Gain, Information Gain Ratio, Chi-squared test and Principal Com-
ponent Analysis as their computation can be adapted to categorical features, as
the LD ones. Then, the features selected from each technique have been used
as input for two recommendation algorithms based on graph-kernels [6]: entity-
based and path-based. Experimental results showed Information Gain as the
best performing technique1 . Information Gain (IG) is defined as the expected
reduction in entropy occurring when a feature is present versus when it is absent.
For a feature fi , IG is computed as [5]:
X |Iv |
IG(fi ) = E(I) − ∗ E(Iv )
|I|
v∈dom(fi )

where E(I) is the value of the entropy of the data, Iv is the number of items
in which the feature fi (e.g. starring for movies) has a value equal to v (e.g. Al
Pacino in the movie domain), and E(Iv ) is the entropy computed on data where
the feature fi assumes value v. The IG of a feature fi is higher as the lower is
the value of the entropy E(Iv ). Features are ranked according to their IG and
the top-k ones are returned.

Schema summarization for feature selection. Linked Data summarization is the
process of extracting a summary of an input linked data set, such that this
summary is smaller (in size) than the input data, but retains information use-
ful for certain tasks. Relevance-oriented summaries capture subsets of the input
data sets and/or ontologies. These subsets are estimated to be more relevant
1
The interested reader may refer to https://github.com/sisinflab/SAC2017/FeatureSelection for
results obtained with other feature selection techniques
for the users according to multidimensional relevance criteria [10]. Vocabulary-
oriented summaries describe the usage of vocabularies, e.g., ontologies, used in
a dataset. These summaries are usually defined so as to be complete, i.e., to
provide information about every element of the vocabulary/ontology used in
the data set [9]. Vocabulary-oriented summaries that provide complete descrip-
tions of vocabulary usage may support feature selection by providing relevant
information about every possible feature, i.e., property, in the data set.
In this paper we use summaries produced by a vocabulary-oriented summa-
rization framework named ABSTAT2 . It takes a linked data set and - when
available - one or more ontologies used in this data set as input, and re-
turns a summary. The summary consists in a set of patterns having the form
hC, P, Di, with C and D being types, i.e., concepts or datatypes, and P being
an RDF property. We refer to C and D as source and target types, respectively.
Each pattern hC, P, Di tells that there exist some instance of type C linked
to some instance of type D through the property P . For example, a pattern
hdbo:Film, dbo:starring, dbo:Actori tells that there are instances of dbo:Film
linked to instances of type dbo:Actor through the property dbo:starring in
the data set. The summary is complete for relational assertions in an RDF data
set, i.e., assertions about individuals: for every relational assertion hx, p, yi that
exists in the data set, at least one pattern is generated, i.e., every such assertion
is represented by at least one pattern. The generation of these patterns is based
on explicit typing assertions, e.g., hdbr:Tom Cruise, rdf:type, dbo:Actori or on
implicit typing assertions (for literals), e.g., 1962-01-01xsd:date extracted from
the dataset. Differently from other approaches that also extract vocabulary-
based patterns from linked data sets [4, 3], ABSTAT applies a pattern minimal-
ization technique leveraging the relations between types defined in the ontolo-
gies (when the ontologies are used in the summarization process). Additional
information provided in summaries and of major importance for feature selec-
tion is pattern frequency, which counts the occurrences of patterns in the data
set. For example, hdbo:Film, dbo:starring, dbo:Actori[10662] tells that 10662
instances of dbo:Film are linked to instances of type dbo:Actor through the
property dbo:starring in the data set3 .

3 Evaluation

For evaluating the quality of a recommendation algorithm, given a particular
feature selection technique, we use four metrics, as each one of them measures a
different dimension in the final result. To evaluate recommendation accuracy,
we use Precision and Mean Reciprocal Rank (MRR). While P recision@N is a
metric denoting the fraction of relevant items in the top-N recommendations,
2
ABSTAT summaries for several datasets can be explored at http://abstat.disco.unimib.it:8880/
3
For more details about the summarization process, the impact of minimalization on
the size of extracted summaries, the use of ABSTAT summaries to support data
set understanding, and the services through which summaries are accessible via web
interfaces we refer to [9].
Entity-based Graph kernel Top-K features Precision@10 MRR@10 itemCov@10 aggrEntropy@10

5 0.02327 0.15578 0.54262 8.96
IG 10 0.01734 0.13599 0.90658 10.24
15 0.02055 0.14685 0.91989 10.19

5 0.02035 0.14694 0.54953 9.12
ABSTAT 10 0.01651 0.13705 0.64346 9.42
15 0.02062* 0.13757 0.67417 9.42
Path-based Graph kernel Top-K features Precision@10 MRR@10 itemCov@10 aggrEntropy@10

5 0.02266 0.16248 0.58971 9.12
IG 10 0.01518 0.13221 0.88252 10.26
15 0.01387 0.13069 0.89762 10.25

5 0.02026 0.15310 0.54825 9.13
ABSTAT 10 0.01519 0.13331 0.57461 9.33
15 0.01726* 0.13510* 0.62606 9.46
Table 1. Experimental results using the entity-based and the path-based Graph kernel
recommendation algorithms. In bold the configurations where ABSTAT outperforms
IG (the * symbol indicates that the differences between ABSTAT and the IG baseline
are statistically significant with p − value < 0.001 according to the paired t-test.)

MRR computes the average reciprocal rank of the first relevant recommended
item, and hence results particularly meaningful when users are provided with
few but valuable recommendations (i.e., Top-1 or Top-3)[8]. To evaluate aggre-
gate diversity, we consider catalog coverage, i.e., the percentage of items in the
catalog recommended at least once and aggregate entropy [1]. The former is used
to assess the ability of a system to cover the item catalog, namely to recommend
as many items as possible. While the latter measures the distribution of the rec-
ommendations across all the items, showing whether the recommendations are
concentrated on a few items or are better distributed.
The evaluation of the two feature selection methods, IG and ABSTAT, has
been done via the well-know Movielens 1M dataset. In order to enrich it with
information from Linked Data, we started from a dump of the DBpedia dataset4
and we limited it to the movie domain by linking movies in Movielens dataset
with their corresponding DBpedia entries. Table 1 shows the results for entity-
based and path-based graph kernel algorithms [6], respectively. When selecting
only the first 5 features, the two feature selection methods, IG and ABSTAT,
show good values of accuracy, but lower values of aggregate diversity, especially
in term of coverage. This is not really surprising as with a lower number of
features, the system does not have enough diversified information to select more
items and the effect of the popularity bias is stronger. Increasing the number of
features the value of diversity increases at the expense of the accuracy. However, a
good balance remains between accuracy and diversity thus showing a good trade-
off between the two [2]. The implementation of the recommendation algorithm
presented in this work and all the experimental results are available https:
//github.com/sisinflab/SAC2017.

4
http://downloads.dbpedia.org/2015-10/
References
1. G. Adomavicius and Y. Kwon. Improving aggregate recommendation diversity
using ranking-based techniques. IEEE Transactions on Knowledge and Data En-
gineering, 24(5), May 2012.
2. P. Castells, N. J. Hurley, and S. Vargas. Novelty and diversity in recommender
systems. In Recommender Systems Handbook. Springer US, Boston, MA, 2015.
3. T. Gottron, M. Knauf, A. Scherp, and J. Schaible. ELLIS: interactive exploration
of linked data on the level of induced schema patterns. In Proceedings of the 2nd
International Workshop on Summarizing and Presenting Entities and Ontologies.,
CEUR Workshop Proceedings, 2016.
4. N. Mihindukulasooriya, M. Poveda-Villalón, R. Garcı́a-Castro, and A. Gómez-
Pérez. Loupe - an online tool for inspecting datasets in the linked data cloud.
In Proceedings of the ISWC 2015 Posters & Demonstrations Track, CEUR Work-
shop Proceedings, 2015.
5. C. Musto, P. Lops, P. Basile, M. de Gemmis, and G. Semeraro. Semantics-aware
graph-based recommender systems exploiting linked open data. In Proceedings of
the 24th Conference on User Modeling Adaptation and Personalization, UMAP
2016, 2016.
6. V. C. Ostuni, S. Oramas, T. Di Noia, X. Serra, and E. Di Sciascio. Sound and
music recommendation with knowledge graphs. ACM Transactions on Intelligent
Systems and Technology (TIST), 2016.
7. A. Ragone, P. Tomeo, C. Magarelli, T. Di Noia, M. Palmonari, A. Maurino, and
E. Di Sciascio. Schema-summarization in linked-data-based feature selection for
recommender systems. In Proceedings of the Symposium on Applied Computing,
SAC ’17, pages 330–335. ACM, 2017.
8. Y. Shi, A. Karatzoglou, L. Baltrunas, M. Larson, N. Oliver, and A. Hanjalic. Climf:
learning to maximize reciprocal rank with collaborative less-is-more filtering. In
Proceedings of the sixth ACM conference on Recommender systems. ACM, 2012.
9. B. Spahiu, R. Porrini, M. Palmonari, A. Rula, and A. Maurino. ABSTAT: ontology-
driven linked data summaries with pattern minimalization. In Proceedings of the
2nd International Workshop on Summarizing and Presenting Entities and On-
tologies (SumPre 2016) co-located with ESWC., volume 1605 of CEUR Workshop
Proceedings. CEUR-WS.org, 2016.
10. G. Troullinou, H. Kondylakis, E. Daskalaki, and D. Plexousakis. RDF Digest:
Efficient Summarization of RDF/S KBs. In ESWC, 2015.