=Paper=
{{Paper
|id=Vol-2903/IUI21WS-SOCIALIZE-7
|storemode=property
|title=Information Extraction for Inclusive Recommender Systems
|pdfUrl=https://ceur-ws.org/Vol-2903/IUI21WS-SOCIALIZE-7.pdf
|volume=Vol-2903
|authors=Noemi Mauro,Liliana Ardissono,Stefano Cocomazzi,Federica Cena
|dblpUrl=https://dblp.org/rec/conf/iui/MauroACC21
}}
==Information Extraction for Inclusive Recommender Systems==
https://ceur-ws.org/Vol-2903/IUI21WS-SOCIALIZE-7.pdf
Information Extraction for Inclusive
Recommender Systems
Noemi Mauroa , Liliana Ardissonoa , Stefano Cocomazzia and
Federica Cenaa
a Computer Science Department, University of Torino, Corso Svizzera 185, Torino, I-10149, Italy
Abstract
Inclusive recommender systems should take both user preferences and the compatibility of items with
the user into account in order to generate suggestions that can be appreciated and smoothly experienced
at the same time. For instance, considering people in the Autism Spectrum Disorder, the sensory features
of a place that is potentially interesting to the user are important to predict whether it might make
her/him uncomfortable when visiting it. However, information about users’ experience with items can
hardly be found in the metadata provided by online geographic sources.
In order to address this issue, we suggest to retrieve it from the consumer feedback collected by
location-based services that publish item reviews. This type of feedback represents a sustainable in-
formation source because it is supported by people through a continuous reviewing activity. Thus, it
deserves special attention as a potential data source. In this paper, we outline how this type of informa-
tion can be retrieved and we discuss its benefits to Top-N recommendation of Points of Interest.
Keywords
Recommender Systems, Geographic Information Systems, People with Autism
1. Introduction ically impaired. Moreover, if the user is in the
Autism Spectrum Disorder (ASD), the sen-
In the development of inclusive recom- sory features of places have to be taken into
mender systems, multiple factors have to be account to suggest PoIs that are compatible
taken into account to support a positive user with her/his aversions. However, physical
experience with the suggested items. For in- and sensory features of items are only a part
stance, in Points of Interest (PoIs) recommen- of a broader scope of characteristics that can
dation, accessibility issues, such as architec- negatively influence user experience. For in-
tonic barriers, should be considered to avoid stance, cultural aspects might impose con-
imposing extra fatigue on the user, or mak- straints on garment styles to be considered
ing it hard to reach the place, if (s)he is phys- in order to avoid offending the user with pro-
posals that (s)he cannot accept.
Joint Proceedings of the ACM IUI 2021 Workshops, April These examples support the idea that both
13–17, 2021, College Station, USA user preferences, and the compatibility of
" noemi.mauro@unito.it (N. Mauro); items with the user are key to the suggestion
liliana.ardissono@unito.it (L. Ardissono);
of items that (s)he can like and smoothly ex-
stefano.cocomazzi@edu.unito.it (S. Cocomazzi);
federica.cena@unito.it (F. Cena) perience at the same time. Different compati-
0000-0001-8234-3266 (N. Mauro); bility aspects might be modeled within a rec-
0000-0002-1339-4243 (L. Ardissono); ommender system, depending on its goals,
0000-0003-3481-3360 (F. Cena)
© 2021 Copyright for this paper by its authors. Use permit- such as impairments, sensory aversions, cul-
ted under Creative Commons License Attribution 4.0 Inter-
national (CC BY 4.0). tural principles, and so forth. The basic dif-
CEUR
http://ceur-ws.org
CEUR Workshop Proceedings ference with respect to preference modeling
(CEUR-WS.org)
Workshop ISSN 1613-0073
Proceedings
is that the system should not assess whether 2. Background and related
the user likes more or less a property of an
item, but if that property can cause any dis-
work
comfort, or difficulties, to her/him. Thus, for As discussed by Ghose and Ipeirotis in [4],
instance, a single, totally incompatible fea- online reviews are a precious source of in-
ture might make an item unsuitable for the formation about products and services be-
user, even though its other features would cause they describe previous consumers’ ex-
satisfy her/him very well. perience with items. Notice also that, as re-
A few mobile guides propose models for views are voluntarily provided by people all
the evaluation of items compatibility with over the world, and they are continuously
users in personalized recommendation. For uploaded, they represent an ideally unlimited
instance, INTRIGUE [1] focuses on physical source of up-to-date information about items
accessibility of items, while PIUMA [2, 3] in- that can be used to feed a recommender sys-
vestigates their compliance with the sensory tem.
aversions of people. In both works, the col- A lot of work has been carried out to ex-
lection of information about items support- tract relevant data from online reviews with
ing compatibility evaluation is a problematic the purpose of personalizing recommenda-
task. Specifically, while geographic informa- tion to the individual user [5, 6], or to as-
tion sources provide some accessibility data sess review helpfulness [4, 7, 8, 9, 10]. More-
(e.g., wheelchair support), they typically of- over, a parallel research thread applies opin-
fer standard types of information which can ion mining to identify pros and cons of items,
hardly represent the user experience with as observed by consumers, with the aim of
items in a complete way. Moreover, metadata highlighting aspects that can be improved or
do not always reflect real user experience. promoted. For instance, see [11], [12], and
For instance, even though a hotel claims that [13]. However, those works rely on statistical
it offers Wi-Fi, the quality of internet ac- analyses of text and they focus on the most
cess can only be measured when visiting the frequently reported aspects of items, such as
place. the price and cleanliness of a hotel, or the
In order to address this issue, we want quality of the food served by a restaurant.
to investigate the descriptive power of con- In contrast, compatibility evaluation is re-
sumer feedback collected by location-based lated to individual idiosyncrasies. For this
services that publish online item reviews. reason, it should be based on a deep inves-
Specifically, we want to study the extrac- tigation of users’ perceptions of items, re-
tion of data about sensory features from tex- gardless of how many people highlight the
tual comments in order to check whether various issues in their reviews. For instance,
leveraging this type of information in Top- even though a single person points out that a
N recommendation improves the suggestion restaurant is challenging for somebody who
of places. uses a wheelchair because the tables are too
close to each other, this information should
be recorded and taken into account by the
system. For this reason, instead of identi-
fying the main item properties that emerge
from a statistical analysis of consumer feed-
back (bottom-up), we aim to start from the
identification of the types of features that can Then, we plan to use Natural Language Pro-
determine a compatibility problem. Then, cessing tools to extract the occurrences of
we want to search for these features in the such words from the bulk of reviews associ-
reviews (top-down). We hypothesize that ated with each individual item. In this way,
this approach has the advantage that isolated we can build an item profile that specifies, for
opinions can bring useful data to be used each feature, the mean value emerging from
in the cautious type of recommendation we the complete set of occurrences of the asso-
pursue. ciated words.
Firstly, we will focus on sensory features
to use them in compatibility evaluation of
3. Item recommendation PoIs in relation to people with autism. For
this purpose, we are investigating online re-
We plan to exploit the feature values ex-
view repositories providing consumer feed-
tracted from consumer feedback to evaluate
back about PoIs, such as Yelp [14], TripAdvi-
the compatibility of each feature 𝑓 of an item
sor [15], and Google Maps [16]. However, we
𝑖 with a user 𝑢, taking her/his aversions into
will extend our analysis to other types of fea-
account. Following the approach presented
tures, such as those related to physical acces-
in [2], we will combine the compatibility of
sibility, in order to complement the standard
the features of 𝑖 with 𝑢’s preference for the
type of information provided by data-sources
category of 𝑖 to obtain the final score of the
such as OpenStreetMap [17] with the percep-
item. Compatibility and preference informa-
tions of previous visitors.
tion can be integrated in different ways. For
We also plan to combine the extracted
instance, in [2] we proposed to compute an
information with other data sources. For
overall compatibility value 𝑐𝑜𝑚𝑝 for the item
instance, we will consider OpenStreetMap
and to combine it with 𝑢’s preference 𝑝𝑟𝑒𝑓
for metadata provision and possibly crowd-
for the category 𝑐 of 𝑖 (e.g., cinema, park, etc.)
sourcing platforms such as Maps4All [18],
in order to estimate the rating 𝑟̂ of 𝑖:
which support a flexible type of geo-data
𝑟̂ = 𝛼 ∗ 𝑐𝑜𝑚𝑝 + (1 − 𝛼) ∗ 𝑝𝑟𝑒𝑓 (1) mapping. In this way, we will possibly obtain
richer item profiles to be used for person-
where 𝛼 ∈ [0, 1] tunes the influence of com- alized recommendation. We are aware that
patibility and preference information in rat- is hard to collect real, objective sensory fea-
ing estimation. However, other methods can tures of PoIs, since the same place can be per-
be applied, which we plan to investigate. ceived differently from person to person, es-
pecially with notable differences between in-
dividuals with autism or not. However, we
4. Extraction of think that, by merging a large amount of dif-
ferent points of view on the same place, as
compatibility features provided by online reviews, we will be able
from online reviews to obtain an image as similar as possible to
its real characteristics.
In order to support a top-down search of
compatibility features in item reviews, we
plan to identify the words that refer to such
features and to map words to feature values.
Acknowledgments actions on on Knowledge and Data
Engineering 23 (2011) 1498–1512. URL:
We thank Claudio Mattutino and the Adult https://doi.org/10.1109/TKDE.2010.188.
Autism Center of the City of Torino for their doi:10.1109/TKDE.2010.188.
support to this work. [5] M. Hernández-Rubio, I. Cantador,
A. Bellogín, A comparative analy-
sis of recommender systems based
References on item aspect opinions extracted
[1] L. Ardissono, A. Goy, G. Petrone, from user reviews, User Mod-
M. Segnan, P. Torasso, INTRIGUE: eling and User-Adapted Interac-
personalized recommendation of tion 29 (2019) 381–441. URL: https:
tourist attractions for desktop and //doi.org/10.1007/s11257-018-9214-9.
handset devices, Applied Artificial doi:10.1007/s11257-018-9214-9.
Intelligence, Special Issue on Artificial [6] L. Chen, G. Chen, F. Wang, Rec-
Intelligence for Cultural Heritage and ommender systems based on user
Digital Libraries 17 (2003) 687–714. reviews: the state of the art, User
URL: https://www.tandfonline.com/ Modeling and User-Adapted Inter-
doi/abs/10.1080/713827254. doi:https: action 25 (2015) 99–154. URL: https:
//doi.org/10.1080/713827254.
//doi.org/10.1007/s11257-015-9155-5.
[2] N. Mauro, L. Ardissono, F. Cena, Per- doi:10.1007/s11257-015-9155-5.
sonalized recommendation of PoIs to [7] R. Dong, M. Schaal, M. P. O’Mahony,
people with autism, in: Proceedings B. Smyth, Topic extraction from on-
of the 28th ACM Conference on User line reviews for classification and rec-
Modeling, Adaptation and Personaliza- ommendation, in: Proceedings of the
tion, UMAP’20, ACM, New York, NY, Twenty-Third International Joint Con-
USA, 2020, pp. 163–172. URL: https://dl. ference on Artificial Intelligence, IJCAI
acm.org/doi/10.1145/3340631.3394845. ’13, AAAI Press, 2013, p. 1310–1316.
doi:10.1145/3340631.3394845. [8] H. Hong, D. Xu, G. Wang, W. Fan, Un-
[3] A. Rapp, F. Cena, C. Mattutino, derstanding the determinants of online
G. Boella, C. Schifanella, R. Keller, review helpfulness, Decision Support
S. Brighenti, Designing an urban Systems 102 (2017) 1–11. URL: https:
support for autism, in: Proceedings //doi.org/10.1016/j.dss.2017.06.007.
of the 21st International Confer- doi:10.1016/j.dss.2017.06.007.
ence on Human-Computer Inter- [9] M. Siering, J. Muntermann, B. Ra-
action with Mobile Devices and jagopalan, Explaining and pre-
Services, MobileHCI 2019, Taipei, dicting online review helpfulness:
Taiwan, October 1-4, 2019, ACM, the role of content and reviewer-
2019, pp. 43:1–43:6. URL: https: related signals, Decision Sup-
//doi.org/10.1145/3338286.3344390. port Systems 108 (2018) 1 – 12.
doi:10.1145/3338286.3344390. URL: http://www.sciencedirect.com/
[4] A. Ghose, P. G. Ipeirotis, Estimating science/article/pii/S0167923618300149.
the helpfulness and economic impact doi:10.1016/j.dss.2018.01.004.
of product reviews: mining text and [10] N. Mauro, L. Ardissono, G. Petrone,
reviewer characteristics, IEEE Trans- User and item-aware estimation of re-
view helpfulness, Information Process-
ing & Management 58 (2021) 102434.
URL: http://www.sciencedirect.com/
science/article/pii/S0306457320309274.
doi:https://doi.org/10.1016/j.
ipm.2020.102434.
[11] J. Qi, Z. Zhang, S. Jeon, Y. Zhou,
Mining customer requirements from
online reviews: A product improve-
ment perspective, Information &
Management 53 (2016) 951 – 963.
URL: http://www.sciencedirect.com/
science/article/pii/S0378720616300581.
doi:10.1016/j.im.2016.06.002.
[12] E. Bilici, Y. Saygın, Why do people (not)
like me?: Mining opinion influencing
factors from reviews, Expert Systems
with Applications 68 (2017) 185 – 195.
URL: http://www.sciencedirect.com/
science/article/pii/S0957417416305322.
doi:10.1016/j.eswa.2016.10.001.
[13] X. Xu, X. Wang, Y. Li, M. Haghighi,
Business intelligence in online
customer textual reviews: under-
standing consumer perceptions and
influential factors, International
Journal of Information Manage-
ment 37 (2017) 673 – 683. URL:
http://www.sciencedirect.com/science/
article/pii/S0268401217301378. doi:10.
1016/j.ijinfomgt.2017.06.004.
[14] Yelp, Yelp, 2019. https://www.yelp.com.
[15] TripAdvisor, Tripadvisor, 2017. https:
//www.tripadvisor.it/.
[16] Google, Google maps, 2019. https:
//www.google.com/maps.
[17] OpenStreetMap Contribu-
tors, Openstreetmap, 2017.
https://www.openstreetmap.org.
[18] Social Computing, Firstlife - maps4all,
2020. https://maps4all.firstlife.org.