ON THE NEED FOR INCENTIVES TO SUPPORT

PERSONALIZATION SYSTEMS

Turning Users into Active Providers of Contents and Metadata

Martín López-Nores, José J. Pazos-Arias, Jorge García-Duque

Yolanda Blanco-Fernández, Alberto Gil-Solla and Manuel Ramos-Cabrer

Department of Telematics Engineering, University of Vigo, Spain

Keywords:

Personalization, metadata, incentive schemes, social networks.

Abstract:

Research in personalization systems has made enormous progress in the last few years. However, the phe-

nomenon of information overload is taking the state of the art to a dead end, due to the lack of metadata to

describe the growing number of available contents. In this position paper, we take a look at the problem and

suggest a research roadmap to ﬁnd a way out, working on the idea of providing incentives to the end users to

become active providers of contents and metadata.

1 INTRODUCTION

In recent years, we havewitnessed the developmentof

new communication technologies and a rapid growth

in the amount of information available. In this sce-

nario, the users would be expected to beneﬁt greatly

from a wide range of services delivering news, enter-

tainment, education, health care advice, commercial

facilities and so on. However, the current situation

may be better referred to as one of information over-

load, since it frequently happens that the users are

faced with an overwhelming amount of information.

A similar situation was seen in the 1990s with the

exponential growth of the Internet, that made users

feel disoriented among the myriad of contents avail-

able through their PCs. This gave birth to the suc-

cessful search engines (e.g. Google and Yahoo), that

would retrieve relevant web pages in response to user-

entered queries. Nonetheless, the advent of new de-

vices (Digital TV set-top-boxes, mobile phones, me-

dia players, ...) brings into scene consumption habits

that render the search engine paradigm insufﬁcient.

For various reasons, it is no longer realistic to think

that users will bother to visit a site, enter queries de-

scribing what they want, and select particular contents

from a list. In response to that, the scientiﬁc commu-

nity is devoting huge efforts to the design and pro-

vision of personalized information services (Ardis-

sono et al., 2004; Chorianopoulos, 2008; Im and Park,

2007), with a new paradigm of recommender systems

proactively selecting the contents that best match the

interests and needs of each individual at any time (Lee

and Yang, 2003; Hung, 2005; Cao and Li, 2007;

Tsunoda and Hoshino, 2008).

2 THE PROBLEM

Recommender systems work by matching user pro-

ﬁles against metadata that describe the available con-

tents (Adomavicius and Tuzhilin, 2005; Burke, 2002;

Lim et al., 2008). The point we want to highlight is

that, precisely due to the problem of information over-

load, content providers are already incapable of pro-

ducing metadata to characterize the growing amount

of material; they might provide some metadata for

the new contents they produce, but certainly not for

the many ones that already exist, neither for user-

contributed ones. As long as automatic markup or

nearby techniques (Guio and Jay Kuo, 2001; Wang

et al., 2004) are not yet mature enough to ﬁll in this

gap, the peril lurks in the medium term that personal-

ization engines may collapse for lack of input.

Echoing the ongoing revolution of the Web 2.0,

the solution to this problem can only come from the

end users, who should be engaged not only as content

producers, but also as producers of metadata to an-

notate resources. This is already happening to some

extent in social tagging Internet sites like del.icio.us,

CiteULike and Technorati, but the amount of informa-

tion so gathered is still several degrees of magnitude

332

López-Nores M., J. Pazos-Arias J., García-Duque J., Blanco-Fernández Y., Gil-Solla A. and Ramos-Cabrer M. (2008).

ON THE NEED FOR INCENTIVES TO SUPPORT PERSONALIZATION SYSTEMS - Turning Users into Active Providers of Contents and Metadata.

In Proceedings of the International Conference on Signal Processing and Multimedia Applications, pages 332-335

DOI: 10.5220/0001939403320335

 SciTePress

lower than needed for a holistic scenario of personal-

ized services in diverse applications and over multiple

devices. In our opinion, taking the social approach to

a new scale is not just a question of providing suit-

able tools and interfaces for the task in more services

and devices, in the belief that the users will start to

use them just like they started to massively contribute

text, photos and videos to sites like Wikipedia, Flickr

or YouTube. Contrary to providing contents, provid-

ing metadata is an ungratefulactivity, because nobody

sees the metadata: one does not gain visibility to other

users, and so gets no feedback from whoever may be

beneﬁtting from his/her contributions.

3 THE SOLUTION

To bridge the crucial difference between providing

contents and providing metadata, our position is that

it will be mandatory to design incentive schemes to re-

ward the users for any valuable information they pro-

vide (e.g. with coupons for pay-per-view services,

free recharge vouchers for prepaid mobile phones,

hours of premium access to certain contents, dis-

count on broadband connections, technological gad-

gets, etc) and to build the knowledge bases for the rec-

ommender systems in a collaborative fashion. Proba-

bly, the best context for these incentive schemes will

be that of social networks like MySpace or Facebook,

though enhanced with trust and prestige indicators to

promote serious involvement among the contributing

users. Likewise, it will be necessary to research into

how to ensure the traceability of each user’s contribu-

tions, including contents, metadata and even recom-

mendations to others.

In addition, innovations will be needed regarding

the usability of the interfaces offered to the users to

enter information and to interact with the social net-

work, especially knowing that the methods employed

thus far on the Internet (as accessed from personal

computers) are not well suited to the input and presen-

tation capabilities of other devices. Research will be

needed on the design of specialized interfaces able to

tune the incentives offered to the users depending on

the type of information they provide —for instance, to

compensate for the inconvenienceof typing text using

a remote control or a keypad.

Internally, advances must be pursued to work with

novel, rich data structures halfway between ontolo-

gies and folksonomies. In this regard, to enable au-

tomatic processing of the knowledge bases, it will be

necessary to introduce structure in the universeof tags

that may be coming from a social network, but with-

out the traditional well-formedness and consistency

requirements of ontologies to reckon the fact that dif-

ferent viewers may well provide contradictory cate-

gorizations for the same contents or products. This

approach, supported by enhanced data mining tech-

niques (Han and Kamber, 2005), is necessary to ex-

ploit all the knowledge captured in potentially contra-

dictory metadata, with no need to restrict the reason-

ing to consistent subsets as it happens with most of

the works in literature.

The ﬁnal grand topic we envisage relates to the

business models of personalized information systems,

rethinking the relationships among content and ser-

vice providers, network operators, advertisers and

users around the innovations of the incentivized so-

cial network, with special concern for data ownership

and privacyissues. In this point, it is worth noting that

a wealthy ﬂow of information from the users may pro-

vide the foundations for a new framework to perform

audience and market studies in diverse areas of appli-

cation, which will be crucial to support the strategic

actions of the involved stakeholders.

4 A POSSIBLE SCENARIO

The following is a scenario that illustrates the afore-

mentioned innovations in the context of personalized

advertising through Digital TV set-top boxes and mo-

bile devices, including glimpses of the following:

• business models in personalized t-commerce;

• incentives to the TV viewers, graded to the value

of the feedback provided;

• incentives to other users in the consumption

chain;

• indicators for audience and market studies.

On Wednesday evening, channel INCENTV-

SPORTS will broadcast a UEFA Champions League

match between Real Madrid C.F. and A.C. Milan in

pay-per-view mode. To decide what ads to deliver

during the transmission, the channel managers decide

to follow the audience stereotypes of “people who like

football” and “people who like travelling”, so the

head-end is scheduled to deliver material related with

football (sports clothing, merchandising, almanacs,

tickets for upcoming matches, etc) and tourist desti-

nations in Spain and Italy.

It’s Wednesday, and Alice goes to her favourite

bar to watch the game on TV (she prefers this in-

stead of paying e10 to watch it in home). Since

there are be many people in the bar, the TV screen

ON THE NEED FOR INCENTIVES TO SUPPORT PERSONALIZATION SYSTEMS - Turning Users into Active

Providers of Contents and Metadata

333

displays the most relevant ads according to the au-

dience stereotypes, while the individuals can receive

personalized offers in their mobile phones. In a given

moment, Alice presses a YELLOW button in her mo-

bile to indicate that she wants to learn more about

the product currently advertised in the bar’s TV, but

later (RED would mean that she is not at all inter-

ested in that stuff, while GREEN would mean that she

wants to learn more about the offer right now). Fol-

lowing this indication, once the ﬁrst half of the game

is over, Alice’s mobile vibrates to face her with an in-

teractive application that lets her browse t-shirts of

different clubs, prices and so on. Alice could buy

one item using the application, but now she doesn’t

want to. However, thanks to the incentive schemes of

INCENTV-SPORTS, she can still beneﬁt from provid-

ing some information about the offered products. The

more information, the greater the reward:

• By simply providing numerical ratings (from 0 to

10), she collects points that she may exchange for

backgrounds, screensavers and tunes for her mo-

bile phone.

• By tagging products with words from a given vo-

cabulary, she gets points to exchange for products

also advertised on INCENTV-SPORTS.

• By entering new tags to classify products, she wins

discounts to watch the next match at home (e.g.

e0.1 per tag, to a maximum of e2). Textual com-

ments would provide additional discounts of 1e.

Prior to sending it out, Alice can review the data

that will be sent to INCENTV-SPORTS, ﬁnding that

she can supplement what she entered with context in-

formation given by the bar owner (e.g. describing the

bar’s atmosphere or common likings of its clients).

Thus, the bar owner can also get rewards, propor-

tional to the amount of feedback gathered from his

clients and the amount of products they buy. In fact,

next week he will be able to offer a new match for free.

On Friday, the INCENTV-SPORTS managersstart

processing all the feedback gathered from the football

match. Firstly, it is noticeable that there has been very

little activity around tourist destinations ads. This in-

dicates that a football match is not a suitable place to

advertise travels, so the stereotype “people who like

travelling” loses relevance with regard to football. In

contrast, there has been much activity around sports

stuff, measuring great disparity in the viewers’ reac-

tions to certain items that ﬁtted well within the stereo-

type of “people who like football”. In response to

these observations, an analyst suggests to specialize

the stereotype, introducing subclasses of “supporters

of A.C. Milan” and “supporters of Inter Milan”. As

a result, during the next Champions League round,

the supporters of A.C. Milan are not faced with mer-

chandising of Inter Milan and vice versa. Everyone

is happier with the publicity received and an increase

occurs in the sales ﬁgures.

5 CONCLUSIONS

With the growing amount of contents available

through different media, personalized information

services face a risk of starvation due to lack of meta-

data to reason about. In order to solve this problem,

we have argued that it is necessary to turn the users

into active providers of metadata, for which they must

be given suitable incentives in a social network. Real-

izing this view will require much research in incentive

schemes, business models, user interfaces, knowledge

modeling and data mining techniques, among others.

ACKNOWLEDGEMENTS

The visions presented herein will drive the authors’

research within the Ministerio de Educación y Cien-

cia (Gobierno de España) project TSI2007-61599,

also supported by the Consellería de Educación e Or-

denación Universitaria (Xunta de Galicia) incentives

ﬁle 2007/000016-0, and by the Consellería de In-

novación, Industria e Comercio (Xunta de Galicia)

project PGIDIT05PXIC32204PN.

REFERENCES

Adomavicius, G. and Tuzhilin, A. (2005). Towards the

next generation of recommender systems: A survey

of the state-of-the-art and possible extensions. IEEE

Transactions on Knowledge and Data Engineering,

17(6):739–749.

Ardissono, L., Gena, C., Torasso, P., Bellifemine, F.,

Chiarotto, A., Diﬁno, A., and Negro, B. (2004). User

modeling and recommendation techniques for person-

alized Electronic Program Guides. In Personalized

Digital Television. Targeting programs to individual

users. Kluwer Academic Publishers.

Burke, R. (2002). Hybrid recommender systems: Survey

and experiments. User Modeling and User-Adapted

Interaction, 12(4):331–370.

Cao, Y. and Li, Y. (2007). An intelligent fuzzy-based rec-

ommendation system for consumer electronic prod-

ucts. Expert Systems with Applications, 33(1):230–

240.

Chorianopoulos, K. (2008). Personalized and mobile digital

TV applications. Multimedia Tools & Applications,

36(1):1–10.

SIGMAP 2008 - International Conference on Signal Processing and Multimedia Applications

334

Guio, J. and Jay Kuo, C. C. (2001). Semantic video object

segmentation for content-based multimedia applica-

tions. Springer.

Han, J. and Kamber, M. (2005). Data mining: Concepts

and Techniques. Morgan Kaufmann.

Hung, L. (2005). A personalized recommendation system

based on product taxonomy for one-to-one marketing

online. Expert Systems with Applications, 29:383–

392.

Im, K. H. and Park, S. C. (2007). Case-based reasoning and

neural network based expert system for personaliza-

tion. Expert Systems with Applications, 32(1):77–85.

Lee, W.-P. and Yang, T.-H. (2003). Personalizing informa-

tion appliances: A multi-agent framework for TV pro-

gramme recommendations. Expert Systems with Ap-

plications, 25(3):331–341.

Lim, J., Kim, M., Lee, B., Kim, M., Lee, H., and Lee, H.-

K. (2008). A target advertisement system based on

TV viewer’s proﬁle reasoning. Multimedia Tools &

Applications, 36(1-2):11–35.

Tsunoda, T. and Hoshino, M. (2008). Automatic metadata

expansion and indirect collaborative ﬁltering for TV

program recommendation system. Multimedia Tools

& Applications, 36(1-2):37–54.

Wang, J., Xu, Y., Shum, H.-Y., and Cohen, M. F. (2004).

Video tooning. ACM Transactions on Graphics,

23(3):574–583.

ON THE NEED FOR INCENTIVES TO SUPPORT PERSONALIZATION SYSTEMS - Turning Users into Active

Providers of Contents and Metadata

335