=Paper=
{{Paper
|id=Vol-2030/HAICTA_2017_paper38
|storemode=property
|title=Adopting ICT Tools by Farm in Lucania Region
|pdfUrl=https://ceur-ws.org/Vol-2030/HAICTA_2017_paper38.pdf
|volume=Vol-2030
|authors=Gianluigi De Pascale,Piermichele La Sala,Nicola Faccilongo,Claudio Zaza
|dblpUrl=https://dblp.org/rec/conf/haicta/PascaleSFZ17
}}
==Adopting ICT Tools by Farm in Lucania Region==
https://ceur-ws.org/Vol-2030/HAICTA_2017_paper38.pdf
Adopting ICT Tools by Farms in Lucania Region
De Pascale Gianluigi1*, La Sala Piermichele1, Faccilongo Nicola1, Zaza Claudio1
1
Department of Economics, University of Foggia, Italy,
e-mail: gianluigi.depascale@unifg.it
Abstract. This study aims at investigating how Lucania' farms cluster
according to the level of innovation adopted. It was used a questionnaire for
asking if farms adopts ICTs tools and, in case, what type they involved in
managing and/or production processes. It has been done a cluster analysis on
collected data. Results show that, using k-means clustering method, appear two
clusters: innovators, remaining groups. While, using boxplot representation,
clustered three groups: innovators, early adopters and laggards. Results will be
exploited for identifying good practices in terms of smart devices adopted,
within the H2020 project “Short Supply Chain Knowledge and Innovation
Network - SKIN”.
Keywords: ICT, Clustering Analysis, Lucania, Farms
1 Introduction
Agriculture is a field very suffering low efficiency in carrying out its core activities
due to many reasons coming from quick scenery changes. Such changes have been
fostered by new digital technologies. They appear in integrated system named Farm
Management Information System (FMIS). Nowadays, the general addresses to lead
the growth in Europe come from European Commission (EC). In fact, being able to
cope daily problems means to be able to engage synergies among tangible and
intangible resources inspired by arisen studies on such issues. They set out that smart
growth can be put in practice sharing knowledge and adopting innovations (Contò et
al., 2015). Reducing distances among the available resources and accelerate the
access to them. In addition, it has been stated that is necessary to be in conformity
with ecosystem needs (Debackere et al., 2014). To this end, Information and
Communication Technologies (ICTs) tools play an important role for achieving
mentioned goals. Since ecosystem is quite a lot dynamic and changes frequently
occur, it is complex to manage all data emerging by daily activities and to take under
control the scenario evolutions. Such problems are much more evident in small
medium enterprises (SMEs) where information flows are often stressed by the lack of
capabilities to access to ICT innovations (Contò et al., 2015). These tools allow
farmers reducing asymmetric information, being the main cause of moral hazard and
adverse selection mostly affecting firms operating in international markets. The
information management, in turn, influences internal and external actions (Bian et al.,
315
2016). Hence, the highest concern has to be that of making accessible tools useful to
reduce the information gap. The benefits will affect transactional costs. In fact, it is
possible to find many farms that got economic improvements adopting such
technologies (Deichmann et al., 2016). The demonstration of such claims comes
from developing countries that show high level of growth. To this end, the World
Bank arose data in his yearly report dating back to 2016. The report shows that firms
in poor countries adopt digital technologies getting to be much more
competitiveness, whilst maintaining a low profile in international markets due to the
lack of appropriate skillful and infrastructures as well (World Bank, 2016). On the
other hand, scholars have been arising complex management systems that bring
together all elements making farms up. The hardest challenge is to guarantee a right
resources coordination and employment in long period, in order to attain the general
goal of adding sustainable value to the stakeholders.
This article is structured in different sections, they are organized as follows: the
second section (ii) proposes a literature review; the third section (iii) shows the
outline of the questionnaire used for collecting data farms; the fourth section (iv)
explains the analysis method used for processing data for extracting information.
Finally, there are summarized the results and provided their discussion before the
conclusions.
2 Literature Review
Over the years, many changes have been occurring with the advent of ICTs
technologies, affecting, in particular, the farms efficiency. Scholars have taken this
opportunity to study deepen the impacts of such extraordinary evolutions. The first
step to better understand following consideration on ICTs farms tools, is related to
the reason making necessary to implement innovation processes and knowledge
uptake. It is based on the pace of the cost level showed by farms so far. Nowadays,
European and national policies address organizations to realize a cost reduction
through the adoption of smart devices, being in line with the industry 4.0 topics. To
this extent, world organizations as the World Bank pursues in broadcasting analysed
data showing how costs decrease (World Bank, 2016) by introducing ICT tools for
managing the growing complexity activities, due to the complicated competition and
vice versa (Jain et al., 2011; Chen et al., 2012; Lee and Yang, 2013). Deichmann et
al. (2016) explain what type of problems preventing the digital devices adoption in
developing countries. Obviously, there are many countries divided in different areas,
several ones are into prominent growing processes, and others suffer the absence of
capabilities to acting growth. It depends on many factors. The significant ones
concerns the slowness in reforming business regulations and the skills development
system. They stress the idea that building efficient information system is the key for
triggering sustainable growth in long period. A signal of farms (and more in general
firms) efficiency is related to the impact of smart tools on the amount of the
production, which measure the total factor productivity growth (OECD, 2013).
Within the farms dynamisms, Diedern et al. (2003) distinguished between innovators,
early adopters and laggards. These three categories can represent the farmers profile
316
appearing in European areas. Measuring innovation level within the farm has taken
to different assessing methods. Another classification can be carried out with a
matrix that makes differentiation between innovations as major, intermediate or
minor based on their technological advancement. The matrix shows an innovation
numerical index standing for the innovation level of each farm (Ariza et al., 2013).
Many others studies have explained the agricultural innovations through the
agricultural technologies uses (Dimara and Skuras, 2003; Sauer and Zberman, 2012;
Stefanides and Tauer, 1999). On the other hand, their study issued that structural
characteristics, such as farm size, utilized agricultural areas and age of the farmer,
reflect the attitude and the willing to choose to undertake and pursue innovating
processes. According to these assumptions, they distinguish between innovators and
remaining groups. The literature breakdown also involves a study using more
complex index. This index not considers the adoption of a single technology.
Conversely, it focuses on combined factors defining the innovations. The complexity
comes due to the variables taken into account not only focalize on tools and
equipment (Chen et al., 2014; Esmeijer at al., 2015) for carrying out farms activities
(e.g. ICT tools, tractors, etc.), but also on primary productive factors (e.g. seeds).
Therefore, it brings together different elements combining the effects on the farm
results and showing the benefits from emerging synergies. However, the latter is not
the case of this paper. This study used several of those simple methods for evaluating
the results in terms of innovation level. The first stage identifies the groups through
a clustering analysis and such analysis reveals to exist two main groups (as Dimara
and Skuras, 2003; Sauer and Zberman, 2012; Stefanides and Tauer, 1999 conclusions
revealed): innovators and remaining groups. The next step sheds a light on four
variables mined from the dataset and organized in a boxplot. This examination
highlights that the groups are three groups, complying with the insights coming from
Diedern (2013).
3 Data Collection Method
The questionnaire is composed by twenty-two Questions (Qs) in total. The survey
is divided in two parts: i) General Information, starting from the Q1 to the Q7,
regarding the general aspects of the farms involved in the survey. Based on the Q7
reply, regarding the use of ICT tools, the questionnaire foresees the second section
dedicated to the ii) Farms using ICT (from Q8 to Q22), or it ends in case of negative
answer. In the second section there are set of Qs dedicated to analyse what are the
most used ICT tools applied to the farm management and the impact that these
technologies could have on the decrement of agronomic input and manpower
employed and on the production increasing.
Following the Qs are described:
Q1) Legal status: possible answers (partnership; capital company; others).
Q2) Time of Constitution: possible answers (less than five years; between five and
ten years; more than ten years).
Q3) Farmer’s Age: possible answers (less than thirty-five years; between thirty-
five and fifty years; more than fifty years).
317
Q4) Utilized Agriculture Area (UAA): possible answers (less than ten hectares;
between ten and fifty hectares; more than fifty hectares).
Q5) Crop Type: possible answers (Tree crops, herbaceous crops, mixture crops).
Q6) Income: possible answers (between 0 and 50000€; between 50001 and
120000€; between 120001 and 250000€; between 250001 and 500000€; between
500001 and 1000000€; more than 1000000€).
Q7) Do you use ICT tools? Possible answers (yes or no). If the reply is positive,
the farmer answers the demands from Q8 to Q22, in contrary the questionnaire ends.
Q8) What type of Management Tools do you use? Possible answers (none; tools
for Farm’s notebook; tools for warehouses’ management; tools for management of
balance sheet; tools for management of invoicing; Enterprise Resource
Management; others). Multiple answers are allowed.
Q9) What type of Software for Data Management do you use? Possible answers
(none; software for data storing; software for market analysis; Decision Support
System software; software to analyse the costs; others). Multiple answers are
allowed.
Q10) Do you use tools for Precision Agriculture? Possible answers (yes or no). If
the reply is positive, the farmer answers the demand Q11, in contrary the Q17.
Q11) Do you use environmental sensors? Possible answers (yes or no). If the reply
is positive, the farmer answers the demand Q12, in contrary the Q13.
Q12) Why do you use environmental sensors? Possible answers (Fertilization,
Phytosanitary treatments, Weeding, Irrigation, Sowing, Soil management). Multiple
answers are allowed.
Q13) Do you use Unmanned Aerial Vehicle (UAV or drones)? Possible answers
(yes or no). If the reply is positive, the farmer answers the demand Q14, in contrary
the Q15.
Q14) Why do you use UAV? Possible answers (Fertilization, Phytosanitary
treatments, Weeding, Irrigation, Sowing, Soil management). Multiple answers are
allowed.
Q15) Do you use Satellite Data? Possible answers (yes or no). If the reply is
positive, the farmer answers the demand Q16, in contrary the Q17.
Q16) Why do you use Satellite Data? Possible answers (Fertilization,
Phytosanitary treatments, Weeding, Irrigation, Sowing, Soil management). Multiple
answers are allowed.
Q17) Do you use External Data Sources? Possible answers (yes or no). If the reply
is positive, the farmer answers the demand Q18, in contrary the Q19.
Q18) What types of Data do you research? Possible answers (Agro-
Meteorological, Market, Legal aspects, Phytosanitary bulletin, Others). Multiple
answers are allowed.
Q19) What type of tools do you think is the most useful? Possible answers
(External Data Sources, Enterprise Resource Planning, Software for Data
Management, Precision Agriculture tools).
Q20) Since you started to use ICT tools, do you have detected a reduction in the
use of agronomic inputs (pesticides, fertilizers, water, etc.)? To what extent? Possible
answers (None; between 0 and 5%; between 6 and 10%; between 11 and 20%; more
than 20%).
318
Q21) Since you started to use ICT tools, do you have detected a reduction of
employed manpower? To what extent? Possible answers (None; between 0 and 5%;
between 6 and 10%; between 11 and 20%; more than 20%).
Q22) Since you started to use ICT tools, do you have detected an increment of
production? To what extent? Possible answers (None; between 0 and 5%; between 6
and 10%; between 11 and 20%; more than 20%).
For the cluster analysis presented in the next section a subset of the total variables
was taking into account (Table 1).
Table 1. Variables used in the cluster analysis. The code is associated to a single answer in the
cluster analysis.
Name and
Question Answers Code
abbreviation
Q1 Legal Status Partnership 1
Capital company 2
Q3 Farmer’s Age Less than thirty-five years 1
(Age) Between thirty-five and fifty years 2
More than fifty years 3
Q4 Utilized Less than ten hectares 1
Agriculture Area Between ten and fifty hectares 2
(UAA) More than fifty hectares 3
Q5 Crop Type Tree crops 1
Herbaceous crops 2
Mixture crops 3
Q6 Income Between 0 and 50000€ 1
Between 50001 and 120000€ 2
Between 120001 and 250000€ 3
Between 250001 and 500000€ 4
Between 500001 and 1000000€ 5
Q7 Do you use ICT Yes 1
tools? (ICT) No 2
4 Data Analysis Methods
In this paper, collected data have been analysed using clustering analysis. For
obtaining groups featured by homogeneous parameters, it has been resorted to
considering k-means cluster method. The analysis returned acceptable results setting
two clusters. The choice of selecting two clusters it was possible due to:
• the k-means clustering method can be applied with both supervised and un-
supervised methodology (Wagstaff et al., 2001);
• three clusters not returned acceptable results.
In general, k-means is a method that born as un-supervised. Therefore, processing
machine automatically calculates the least distances, respecting the set threshold
between features (Zhang et al., 1996). The goal aims to evaluate if the distances are
319
such to consider the minimum sum of the squared error (SSE) within each groups
(Likas et al., 2003). The formula of the SSE is the following:
K
2
(1)
SSE = ∑ ∑ x −µ
k =1 ∀xi ∈C k
i k
where Ck is the set of grouped data in cluster k; µk is the vector mean of cluster k.
Using un-supervised method, it was found that the clusters were two. Nevertheless,
for being in line with Diedern et al. (2003), the scope was to find three groups to be
labelled as innovators, early adopters and laggards. The test not achieved the goal
and it was tried with a supervised method setting three clusters. In turn, the test not
succeeded due to the cluster two and three presented identical features (the reason
why un-supervised method returned two clusters). Hence, it has been chosen to apply
a supervised method selecting two clusters. At this stage, the test succeeded and
results were accepted to fulfill the two groups theory (innovators and remaining
groups) issued by Dimara and Skuras (2003), Sauer and Zberman (2012), Stefanides
and Tauer (1999).
Then again, it has been attempted to go through the data, analyzing data through a
boxplot to summarize the frequencies. The analysis comes from intersection of
selected variables. It has chosen to fix variables for creating groups and, to this
extent, UAA and age have been selected. Within each group, it has been investigated
how the presence of ICT tools is bridged to the incomes and legal status.
5 Results and Discussion
The questionnaire shows answering from a sample within a producer organization
(PO) in Lucania region. The respondents are 59. They represent the image of the
region in terms of the typology of farms and, in scale, the farm population
composition in Lucania region.
In this section are exposed the results from the frequency distribution of Q20, Q21
and Q22, and the cluster analysis and the boxplot. The paragraph concludes with
discussion from results, emphasizing the differences of the marks basing insights on
the literature provided in section (ii). The reader finds the frequencies of relevant
selected variables and relative comments as well. The discussion provides insights
for explaining potential barriers, obstructing the ICT tools adoption, to be
investigated with further studies. There are delivered considerations on what the
farmers not ICTs skilled and, in consequence, not adopter, while oriented and
inclined to adopting.
320
a)
b)
c)
Fig. 1. Frequency distribution of answers related to: a) Q20 (Since you started to use ICT
tools, do you have detected a reduction in the use of agronomic inputs (pesticides, fertilizers,
water, etc.)? To what extent?); b) Q21 (Since you started to use ICT tools, do you have
detected a reduction of employed manpower? To what extent?); c) Q22 (Since you started to
use ICT tools, do you have detected an increment of production? To what extent?). None =
None; 0-5% = between 0 and 5%; 6-10% = between 6 and 10%; 11-20% = between 11 and
20%; >20% = more than 20%.
321
In Fig. 1 there are the frequency distributions related to Q20 (a), Q21 (b) and Q22
(c), regarding the 29 farms that use ICT tools. The Q20 and Q21 concerning,
respectively, the reduction of agronomic input, such as fertilizers, pesticides, water,
etc. and employed manpower, recorded by the farmers since they started to adopt
ICT tools. Analysing the Q20 answers, the 17,2% (5 replies) did not notice difference
in the application of the agronomic input, while a little reduction (0-5%) was
detected in the 20,7% (6 replies). The modal value with 11 replies (37,9%) is the
range 6-10%, and the last two intervals, 11-20% and >20%, have collected 5 (17,2%)
and 2 (6,9) replies, respectively. Taking into account the Q21 answer, 6 farmers
(20,7%) did not notice any reduction in the employed manpower related to the use of
ICT tools, while a light decreasing (0-5%) was perceived by 7 farmers (24,1%). Even
in this case, the range 6-10% represents the modal value, with 10 answers
representing the 34,5% of the total, with the last two intervals, 11-20% and >20%,
have collected both 3 replies (10,3%). Finally, the modal value of the frequency
distribution of the Q22 is represented by the first class, with 10 (34,5%) farmers that
did not notice any production increment associated to use of ICT tools. Then the
others range, 0-5%, 6-10%, 11-20% and >20% have collected respectively 6
(20,7%), 7 (24,1%), 2 (6,9%) and 4 (13,8%) answers.
As indicated in the section (iii), the queries aimed to evaluate the number of farms
that adopt ICTs tools and associated evidences from the farm structure and the age of
the farmers. The cluster analysis shows two main clusters characterize as follows
(Table 2):
Table 2. Two emerging clusters after processing data. The result concerns a selected number
of variables from the survey.
CLUSTER
1 2
Legal status 1 1
Crop type 2 3
Age 2 2
UAA 2 2
incomes 2 4
ICT 0 1
The resulting clusters present different features, only concerning three variables:
crop type, incomes and ICT. In general, both clusters appear to consist of farms
established as partnership, with farmers being medium age ranking old (35-50);
farming between 10 and 50 hectares. Differences come from:
• Crop type: the cluster 1 is featured by herbaceous crop, while the cluster 2 is
featured by mixed crop (herbaceous and tree crop);
• Incomes: the cluster 1 gains to the extent between 50.000,00 and 120.000,00
euros, instead the cluster 2 lines up between 250.000,00 and 500.000,00
euros;
• ICT: the cluster 1 is represented by farms not adopting ICT tools, the cluster
2 is featured by farms adopting ICT.
322
Table 3 shows the frequencies within each cluster. There appear the 38 clustered
in the first (1) group and the 21 grouped in the second (2) cluster.
Table 3. Number of cases of each cluster.
Cluster
1 38,000
2 21,000
Valid 59,000
Missing ,000
The major number of farms is concentrated in the first cluster. Although the last
evidence, the cluster analysis points out that, in accordance with Dimara and Skuras,
(2003), Sauer and Zberman (2012), Stefanides and Tauer (1999) asserted, come up
two groups: innovators and remaining groups. The cluster 2, populated by
innovators, registers revenues much more relevant than the 1. The measure
corresponds to the range between 250.000,00 and 500.000,00 euros against the
cluster 1 featuring incomes between 50.000,00 and 120.000,00 euros. Furthermore, it
results that the cluster adopting ICT is characterized by mixed crop. In this regards,
the crop diversification is associated to higher incomes (Di Falco and Zoupanidou,
2017). Taking into consideration that the ones encompassed in the cluster 2 adopt
ICT tools, the result in terms of incomes is significant. When farms adopt ICTs seem
to improve the performance. On the other hand, there is a dependency between those
two variables, though it is not defined the direction: at this stage is not clear what
kind of factors push farms in innovating with ICT. In fact, it can depend on the
achievement of excessive dimension and, due to the increasingly complexity, farms
need to improve the data collection and management phases; otherwise, it can depend
on the need to improve the revenue performance and so, the adoption of ICT tools
cause the incomes increasing. For making clearer the explained point, it has been
done another analysis. Assuming that the variable type of crop is excluded due to it
has been chosen to conduct the analysis not considering such qualitative agronomic
variable. The focus remains on the economic aspects, exploring relations with
economic parameters.
Fig. 2 shows the boxplot where are intersected four different variables, looking for
stratum where farms adopt ICT. Outputs put in evidence that there are three main
groups classified by UAA and age. Firstly, results seem to be in comply with the
conclusions of Diedern et al. (2003). Indeed, the picture features three different
relevant groups that can be summarized as follows:
• Innovators, characterized by age between 18 and 35 and UAA no more 10
hectares;
• Early adopter, mainly featured by age between 35 and 50 and UAA between
11 and 50 hectares;
• Laggards, principally classified by age over 50 and UAA between 35 and 50
hectares.
323
It has been assumed that if the age and the UAA present low value, hence the
ICT adoption positively affects the productivity and, in consequence, the incomes.
The picture displays that the major concentration of the ICT matches with the square
corresponding to age level 1 (18-35) and UAA level 1 (0-10 hectares). This
consideration confirm that younger farmers are much more incentivized and
motivated to resort to ICT tools for managing farming activities (Plechowski, K.
2015). The innovators label is due to the age of the farmers, who, even though the
low profile in terms of utilized lands, got medium-high level of incomes. The data is
also confirmed by the legal status. In fact, the corporations are concentrated within
the innovators group. In this regard, this type of legal status costs more than the one
for partnership, and for sustaining the effort farms need to account sufficient
resources in terms of revenues. This consideration allow answering the question
coming from the previous analysis: in this innovators group ICT seem to push the
incomes. Early adopters are characterized farmers aged between 35 and 50 and
farming lands between 11 and 50 hectares. In that case, the growing of the farm
dimension seems to pull ICT tools for managing the growing data complexity.
Conversely, the group where age corresponds to the level 2 and the UAA to the level
3, even though is represented by a niche of respondents, it is another cluster of
innovators. Therefore, the boxplot clustering analysis goes through the data catching
more details than the first one. As a result, the innovators’ cluster is morphologically
more various than the one emerged from Table 2. Finally, laggards (or not
innovators) are featured by the oldest classified farmers, not interested in introducing
ICT devices in farm processes.
Fig. 2. Boxplot grouping clusters according to age and UAA.
324
6 Conclusions
This article mapped the profile of farms in Lucania region, putting in the spotlight
good practices in terms of smart devices used for improving the efficacy and
efficiency of the farms decisions and daily actions. Practitioners can optimize
processes rising to be considered good practices when they reduce the inputs and
improve the outputs, becoming increasingly revenues. In addition to previous reasons
inspiring this study, the survey was also aimed to show good practices adopted by
Lucania farmers. The goal fits with the objectives of the H2020 (EC, 2015) project
Shot Supply Chain Knowledge Innovation Network (SKIN), granted by European
Commission and started on the 1st November 2016. The project consists in collecting
good practices operating in short food supply chain and involving them in European
building network in order to boost and facilitate knowledge transfer and real
innovation uptake. The metrics indicators referenced within the project activities for
collecting good practices, points out that farms raise to be good practice also if adopt
smart tools, such as ICT tools, for improving the economic, environmental and/or
social sustainability. Grouping collected data in clusters allows identifying the most
significant features qualifying smart organizations. The innovators and early
innovators are ready to get into the network providing their experiences and gaining
from other farms experiences. On the other hand, laggards can benefits after network
will be built and synergies will be engaged. They can align their profile to the
smarter ones. Innovators in terms of ICT adoption are mentioned like the ones able to
promptly fit farm’ activities to the environmental complexity and thence the ICT tool
play an important role in moving to that category due to they return efficiency and
efficacy (if they are rightly implemented). By contrast, laggards, even though the
growing environmental complexity (external factors deflecting the right activities
implementation if not correctly managed) do not adopt solution to make simple
processes. However, innovations come up to be needed in rural and agro-food
transition to allow farms becoming economically sustainable. Such necessity is
implied in the farm size that is mainly medium-small and it reduces the
competitiveness in findings profitable markets. The agricultural shocks are going to
increasingly be frequent due to the market uncertain. ICTs facilitate the information
management and the shock control.
Finally, this study considered a small sample of farms from Metaponto’ area where
are mostly concentrated agricultural activities. It appears, obviously, as weakness.
Nevertheless, it has been tried to look at the composition of the sample interviewing
three different types of farms according to the Crop Type (tree crops, herbaceous and
mixed).
Going back through the study, Lucania region presents different profile according
to the ICT devices uses within farm activities. Looking beyond the simple technology
adoption or not adoption, there appear barriers preventing the innovation access
and/or not enabling a real uptake and opportunities exploitation.
The next step of this work consists in:
• checking the results with a bigger sample;
• going through the barriers and investigating detailed reasons limiting and
constraining ICT adoption.
325
Acknowledges. The result presented in this paper is part of the “SKIN project”
(www.shortfoodchain.eu). This project has received funding from the European
Union’s Horizon 2020 Research and Innovation programme under grant agreement
N. 728055.
References
1. Ariza, C., Rugeles, L., Saavedra, D. and Guatiero, B. (2013) Measuring
innovation in agricultural firms: a methodological approach. The Electronic
Journal of Knowledge Management, 11, p.185-198, www.ejkm.com.
2. Chen, M., Mao, S. and Liu, Y. (2014) Big Data: a survey. Mobile Networks and
Applications, 19, p.171-209.
3. Contò, F., Faccilongo, N. and La Sala, P. (2015) The effects of cloud approach in
short chain administration. International Journal of Agricultural and
Environmental Information Systems, 6, p.19-31.
4. Contò, F., Santini, C., La Sala, P. and Fiore, M. (2016) Reducing information gap
and increasing market orientation in the agribusiness sector: Some evidences
from apulia region. Recent Patents on Food, Nutrition and Agriculture, 8, p.48-
54.
5. Diederen, P., van Meijel, H., Wolters, A. and Bijak, K. (2003) Innovation
adoption in agriculture: innovators, early adopters and laggards. Cahiers
d’économie et sociologie rurales, 67, p.30-50.
6. Di Falco, S. and Zoupanidou, E. (2017) Soil fertility, crop biodiversity, and
farmers’ revenues: Evidence from Italy. Ambio, 46, p.162-172.
7. Esmeijer, J., Bakker, T., Ooms, M. and Kotterink, B. (2015) Data-driven
innovation in agriculture: case study for the OECD KBC2-programme. TNO
2015 R10154.
8. European Commission, EC (2015) Calls for proposals and related activities under
the 2016-17 work programmes under Horizon 2020 — the Framework
Programme for Research and Innovation (2014-20) and under the Research and
Training Programme of the European Atomic Energy Community (2014-18)
complementing Horizon 2020. Official Journal of the European Union
(2015/C399/02).
9. EIP-AGRI, (2015) Focus Group Innovative Short Food Supply Chain
management, Final report.
10. Likas, A., Vlassis, N. and Verbeek, J. J. (2003) The global k-means clustering
algorithm. Pattern recognition, 36, p.451-461.
11. Plechowski, K. (2015) Youth, ICTs and agriculture-exploring how digital tools
and skills influence the motivation of young farmers. eChallenges e-2014
Conference Proceedings, Belfast, 2014, p.1-8.
326
12. Wagstaff, K., Cardie, C., Rogers, S. and Schrödl, S. (2001) Constrained k-means
clustering with background knowledge. Proceedings of the Eighteenth
International Conference on Machine Learning, p. 577-584.
13. Wolfert, S., Ge, L., Verdouw, C. and Bogaardt, M. (2017) Big Data in Smart
Farming – A review. Agricultural Systems, 153, p.69-80.
14. World Bank (2016) World Development Report 2016: Digital Dividends, World
Bank, Washington, D.C.
15. Zhang, T., Ramakrishnan, R. and Livny, M. (1997) BIRCH: A new data
clustering algorithm and its applications. Data Mining and Knowledge Discovery,
1, 2, p.141-182.
327