Categorization and Matching for Drone-based Services

Simona Ibba

, Filippo Eros Pani

and Alberto Buschettu

Department of Electrical and Electronic Engineering, University of Cagliari, Piazza d'Armi, Cagliari, Italy

SulcisDrone, Via Alziator 3, Cagliari, Italy

Keywords: Knowledge Management, Taxonomy, Folksonomy, Categorization of Resources, Matching, Drones.

Abstract: The exchange of supply and demand in drone-based services would benefit from the shared use of an online

platform. Such a platform would need to offer two important opportunities. One is to share the service with

other users, and the other is to receive offers from all providers interacting on the platform, who optimize

their investments by sharing their own resources and technologies. The purpose of this position paper is to

propose a categorization and matching algorithm on which to base this platform. The platform will aim to

facilitate the sharing of services provided through the use of drones. The algorithm will match demand and

offer, and will evolve through the use and the application of all participants (operators, users, lenders of shared

resources). The platform, currently in development, could be the first web-based system in Europe to offer

this model.

1 INTRODUCTION

Over the past few years, the scientific community's

interest in drones has increased. A growing number

of researchers worked on the construction and use of

human-friendly drones. The cost reduction of

electronic components and their miniaturization

allowed the commercialization of drones at reduced

prices and thus contributed to their development

(Floreano and Wood, 2015). These devices are able

to fly into three-dimensional (3D) physical space on

their own, without causing damage to objects and

people.

The drones are widely used for aerial video

footage and found wide application in many civilian

activities: location control, crime monitoring,

observation of quality crops or surveillance of

buildings under construction.

(Boyle, 2015) suggests that the demand for drones

is relevant especially at this time of global economic

crisis and austerity. The market for drones is rapidly

expanding: the venture capital funding to drone

companies in 2014 were over 108 million USD

among 29 different companies, and the estimated

2019 turnover for the Unmanned Aerial Vehicles

(UAV) market worldwide is of 8.4 billion USD.

In our project we want to design and implement

an innovative marketplace for the sale, use or rental

of drones. This application must take into account the

possible users, resources, service providers and

applications connected with the use of drones

(industrial, agricultural, environmental, tourism,

etc.). Each user or resource involved in the platform

belongs to a specific category. In the first analysis we

can identify the following categories: utilities,

drivers, vendors, projects, technology, resources and

services. Next to this more formal categorization, we

think to manage knowledge through integration with

a folksonomy in which users are protagonists in

information control (Mathes, 2004) (Vander Wal,

2007).

In this sense we want to develop the platform with

a high level of innovation and attention for users.

The product is, therefore, the service of an online

platform created with state-of-the-art software

engineering, in compliance with the latest standards

for usability and scalability.

In the platform, demand, offer, cooperation and

purchasing groups can interact thanks to resource

sharing and to an algorithm that categorizes and

matches demand and offer.

The paper is structured as follows: Section Two

shows an overview of resource categorization, while

in Section Three we propose our approach. In Section

Four we describe the platform we propose for

matching demand and offer on drone-based services.

Lastly, Section Five includes the conclusions and

some reasoning about our work.

Ibba, S., Pani, F. and buschettu, A.

Categorization and Matching for Drone-based Services.

DOI: 10.5220/0006070002230227

In Proceedings of the 8th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2016) - Volume 3: KMIS, pages 223-227

ISBN: 978-989-758-203-5

223

2 RELATED WORK

Our main goal is to identify the correlations among

users, tags and resources. An efficient approach is

used by (Mika, 2005), extending the basic model of

an ontology with a social dimension that includes

actors, concepts and instances of concepts. The

authors constructed a graph that takes into account the

relationships between tags and users and

simultaneously between tags and resources. Through

network analysis they were then able to infer implicit

categorization of resources.

Even (Koçak et al., 2013) studied some

connection methods of meanings about the collective

market and they developed a mechanism of public

participation to market conventions. They analyzed

data from markets on eBay to discover implications.

An interesting approach can be found in

(Donsbach et al., 2012), showing a user interface

where users can tag items in an electronic catalog and

get recommendations suited to particular tags. The

tags and tag-item assignments created by each user

are stored in association with the user; once a user has

assigned a tag to a number of items, it is possible to

obtain recommendations that are specific to this tag.

These recommendations may be generated in real

time by a service that identifies items that are related

to the items associated with the tag.

In (Tseng et al., 2012) a system which allows a

user to define a relational tag is described. The

relational tag describes how a first item is related to a

second item within an identified qualifying context.

In (Saoud and Kechid, 2016) a new customized

approach integrating a social profile into a distributed

search system is proposed; this approach exploits the

social profile and the different relations between

social entities. This method can enhance a query

expansion, customize and improve both the source

selection and the result merging process in distributed

information retrieval systems.

A new method to categorize mobile apps, a

search-based annotation paradigm influenced by

machine learning techniques, is the one proposed by

(Chen et al., 2016). This system facilitates the search

of preferred apps.

A work of integrating sentiment information to

address the problem of the personalized tag-based

search in collaborative tagging systems is presented

in (Xie et al., 2016). A market system in fact may be

influenced by recommendations that measure

relevance between user and resource. This

folksonomy-based market should help users find their

preferred products. It is relevant to understand user

behaviors and preferences, and we can employ user-

generated tags to discover their perceptions and

feelings about the resources.

On the other hand, (Sarwar et al., 2000) developed

a recommendation algorithm for an e-commerce

system that connects a formal categorization of

ontologies with user preference. The results of this

approach have allowed an improvement of user

experience.

Lastly, (Heidinger et al., 2010) describe a system

which enables people to share experiences and

recommendations regarding the privacy practices of

data collectors. The basis of this system is a

folksonomy in which a user community tags web sites

on the Internet with privacy-related labels. The

system helps to assess the privacy practices of service

providers, and adapts well to a wide range of privacy

threats.

3 THE PROPOSED APPROACH

The aim of our work is the creation of a categorization

and matching algorithm based on content and tags

entered by users so that it can be easier to share the

variety of services provided by drones. An analysis

process will grasp key concepts from content entered

into the demand/offer side. The process will also

extract and manage qualitative content and relations

included in the information in the system, using

service search methods based on semantic content.

The result will be the creation of models of

concepts and categories that will support users in

finding the information they need, relevantly and

quickly. It will lead to the creation of a knowledge

management system in which all elements will be

identified not only as structured information, but also

according to a folksonomy logic, with users

participating in knowledge building (Ibba and Pani,

2016). The elements will also be available for full-

text search.

The matching algorithm is based on a tagging

system that allows sharers and users to label the

product/services on offer. This algorithm also takes

into consideration the statistical and terminology

relations between the description of a product/service

by a sharer and its perception by users or searches. In

all cases, the content is created through the use of tags

or keywords for searches. User-generated content

(UGC) is thus used to improve the presentation of

content in the platform, also by using geolocalization

metadata. In this way, each user will be able to

quickly find the most suitable product/service for

their needs, context and purpose in a complex system.

The algorithm makes it possible to pair the right

KMIS 2016 - 8th International Conference on Knowledge Management and Information Sharing

224

resource with its user. Moreover, a collaborative

recommendation algorithm will be applied to tags and

user-generated content. The aim in doing this would

be to consider the relevance and quality of resources.

The purpose of the algorithm is thus to classify and

filter tag-described resources according to user

profile, usage context and destination.

The algorithm will be created following two main

guidelines:

- the possibility to pre-filter resources by

importance and complexity level. The pre-filtering

configuration parameters can be changed with time as

new content enters the system;

- the needed flexibility to manage

products/services that can be used in several

application fields.

The algorithm will thus be based on some

fundamental parameters:

- the description of resources with content

analysts, which will be needed to identify the

product/service on offer, but also users' opinion on

that resource. The opinions will come from a

collaborative tagging;

- an User Context Profiling, which analyses the

information associated with users and entered during

registration or use of products/services;

- a matching module, calculating the statistical

correspondence of a resource against the needs of a

specific user (only resources above a set threshold are

considered relevant);

- an User classification module, identifying the

similarity among users;

- a Sharer classification module, determining the

similarity among sharers;

- a classification and filtering module to evaluate

statistical and terminology relations between the self-

introduction of a product/service made by a sharer

using tags, and the perception of the same

product/service, made again using tags, by users;

- a Resource Ranking model that introduces users

to the best products/services.

From content published on the platform, it will be

possible to use algorithms to monitor and classify the

appreciation levels of users on any interest macro-

category, and the reception of a specific service.

Moreover, it will be possible to gather information

on the nature of social interactions among users

through the analysis of the social network.

4 PLATFORM FEATURES AND

TARGET MARKET

A wider analysis of the drone industry, including the

professional profiles that use them, casts light on a

market employing tens of thousands of people. The

number of jobs is still to be split among the different

professional fields where drones are used to provide

services.

The market is on the rise thanks to the emerging

needs of a user base that demands innovation, and that

providers try to target through technological and

structural features, such as sensors and software for

the analysis of increasingly specific data, or the

reduction in component size, or again the

development of bespoke applications for agriculture,

environment, and safety. This complex ecosystem is

burdened by legal, practical and technological issues,

which do not facilitate the market. The biggest hurdle

is still the pulverization of providers, which confuse

the users.

In such a context, the platform currently in

development aims to become the first platform in

Europe for services provided through drones.

Connecting to the platform would be all one needs to

have a professional, customized drone service.

The online platforms will allow anyone to find

qualified solutions that fit with the time and budget

requirements in the complex drone industry, where

aerial specialists coexist with professionals coming

from many different fields.

The owners of drones, specialised equipment, or

drone-related services, could showcase their offers in

this channel, which will allow them to:

- acquire new users;

- interact with colleagues and users;

- remarkably increase profits by developing new

offers.

Since the first design phases, we have thought of

placing much care on the system interoperability

feature, a very important requirement for systems

executed in Cloud environments. This non-functional

feature, the use of open-source components and an

Agile approach of XP-programming type, make the

platform become a system extremely susceptible of

upgrade maintenance, which strongly determines the

success of this kind of initiatives.

Only companies with competent in-house

resources that have the right complementary skills for

this kind of activity can create and ensure such a

system.

The platform is based on an optimization /

matching algorithm and on suitable tools and

technologies for evolving the business of the many

Categorization and Matching for Drone-based Services

225

operators and companies that gravitate around the

drone industry. It makes it possible for demand and

offer to meet, further developing an already growing

market. In the platform, demand (all those needing a

drone-related service) and offer (all those providing

drone-related services) interact freely, liven up the

marketplace, share information, and enter self-

referenced profiles.

The service we propose has the aim of managing

this ecosystem, which is experiencing a tumultuous,

disorderly evolution. It will do that thanks to the

insertion of a matching algorithm into a model driven

by the only attractor and developer for demand/offer:

sharing.

Our vision is that the platform will become the

main one in Europe in the shared drone-related

service industry on the long term. Specifically, the

means through which a user can express interest in an

offer, a request or a partnership is the creation of a

“project”. A project is a public advertisement,

whether it be an offer, a request, or a cooperation ad.

It has a title, a description and some other non-

required metadata.

The metadata include the choice of one or more items

from a pre-existing categorization (to classify the ad,

for example as precision agriculture), or from a pre-

existing skill list (to point out the skills needed, or the

skills offered in an ad). There is also a free tagging

system, and advertisements can be geolocalized.

A project is one of the ways in which users create

new data on the platform. The user data created

within the platform are as follows:

- Profile, i.e. data entered in the user profile;

- Project, which could be the offer of a service /

product, a request for a service/product, or a

cooperation advertisement to offer or request a

service with the participation of other users;

- Proposal, that is users' responses to a project (for

example, a quotation for a service request, or an

expression of interest to an offer);

- Rating, through which one user rates the quality

of another user's performance;

- Searches, that is the searches made within the

platform (such as those made with keywords);

- Comments, i.e. how users interact with the

platform, entering text to interact with other users.

All these users could contribute to identify the

preferences of each user.

In the following high-level diagram, the main

modules of the algorithm are shown. They make use

of the user data container (created according to each

user's interactions with the platform) and from a

resource container, formed from some user data (in

particular by the projects) and representing the shared

tools or professional services.

5 CONCLUSIONS

In this position paper we proposed a categorization

and matching algorithm on which to base an online

platform. The platform will aim to facilitate the

sharing of services provided through the use of

Figure 1: Algorithm modules.

KMIS 2016 - 8th International Conference on Knowledge Management and Information Sharing

226

drones. Connecting to the platform will be all one

needs to do to have a professional, bespoke drone

service, customized for any necessity or location.

The platform will make it possible to immediately

find qualified solutions in the complex drone

industry, where aerial specialists coexist with

professionals coming from many different fields.

The target of the system will mainly be the owners

of remotely piloted aircraft systems of any kind,

specialised peripheral devices, and professional

services related to drones. These people could display

their offers through this shared channel. The benefits

to be gained from this would be remarkable, as an

operator could acquire new customers, interact with

colleagues and users, and increase their profits by

developing new offer plans.

Thanks to a wider, more competitive, and more

collaborative offer, aerial services operators will see

an exponential growth in opportunities, and will be

encouraged to continuously propose new solutions.

On the other hand, users could benefit from new

opportunities by saving or by forming purchasing

groups. These actors will be able to use a tool specific

to the drone domain. It is an industry with such

particular characteristics that it needs a bespoke

platform to set offers and create interactions that

would have been ineffective in general demand/offer

matching tools.

REFERENCES

Boyle, M. J. (2015). The race for drones. In Orbis, Vol. 5,

Issue 9.1, pp. 76-94.

Chen, N., Hoi, S., Li, S., Xiao, X. (2016). Mobile App

Tagging. In Proceedings of the Ninth ACM

International Conference on Web Search and Data

Mining. ACM, pp. 63-72.

Donsbach, A., Gregov, A., Few, J. D., Lehman, J. D.,

Brownell, J. T. (2012). Recommendations based on item

tagging activities of users. U.S. Patent No. 8,122,020,

Google Patents.

Floreano, D., Wood, R. J. (2015). Science, technology and

the future of small autonomous drones. In Nature, Vol.

521, Issue 7553, pp. 460-466.

Heidinger, C., Buchmann, E., Böhm, K. (2010).

Collaborative data privacy for the web. In Proceedings

of the 2010 EDBT/ICDT Workshops, ACM.

Ibba, S., Pani, F. E. (2016), The Challenge of Integrating

Instagram with a Taxonomy for Content Management.

Future Internet, 8.2: 16, 2016. DOI: 10.3390/fi8020016.

Koçak, Ö., Hannan, M. T., Hsu, G. (2013). Emergence of

market orders: Audience interaction and avanguard

influence. Organization Studies, Sage Publications.

Mathes, A. (2004). Folksonomies - Cooperative

Classification and Communication Through Shared

Metadata. In: Computer Mediated Communication –

LIS590CMC, Graduate School of Library and

Information Science, University of Illinois

Urbana-Champaign.

Mika, P. (2005). Ontologies are us: A unified model of

social networks and semantics. In International

semantic web conference, Springer Berlin Heidelberg,

pp. 522-536.

Saoud, Z., Kechid, S. (2016). Integrating social profile to

improve the source selection and the result merging

process in distributed information retrieval. In:

Information Sciences, Vol. 336, pp. 115-128.

Sarwar, B., Karypis, G., Konstan, J., & Riedl, J. (2000,

October). Analysis of recommendation algorithms for e-

commerce. In Proceedings of the 2nd ACM conference

on Electronic commerce, ACM, pp. 158-167.

Tseng, W. M., Frank, M. R., Goldstein, M. L. (2012). U.S.

Patent No. 8,103,614, Google Patents.

Unmanned Aerial Vehicle Market Profile and Forecase

(2013). Executive Summary, Teal Group.

Vander Wal, T. (2007). Folksonomy Coinage and

Definition. http://www.vanderwal.net/folksonomy.html

Xie, H., Li, X., Wang, T., Lau, R. Y. K., Wong, T., Chen,

L., Wang, F. L. Li, Q. (2016). Incorporating sentiment

into tag-based user profiles and resource profiles for

personalized search in folksonomy. In Information

Processing & Management, Vol. 52, Issue 1, pp. 61-72.

Categorization and Matching for Drone-based Services

227