Digital Knowledge Ecosystem for Achieving Sustainable Agriculture Production: A Case Study from Sri Lanka

Full text

Digital Knowledge Ecosystem for achieving
Sustainable Agriculture Production: A Case Study
from Sri Lanka
Athula Ginige1, Anusha I. Walisadeera2,3, Tamara Ginige4,5, Lasanthi De Silva3, Pasquale Di Giovanni6, Maneesh
Mathai1, Jeevani Goonetillake3, Gihan Wikramanayake3, Giuliana Vitiello6, Monica Sebillo6, Genoveffa Tortora6,
Deborah Richards5, Ramesh Jain7
1
Western Sydney University, Australia, {A.Ginige, Maneesh}@westernsydney.edu.au,
2
Universaity of Ruhuna, Sri Lanka, waindika@cc.ruh.ac.lk
3
University of Colombo, Sri Lanka, {lnc, jsg, gnw}@ucsc.lk,
4
Australian Catholic University, Australia, Tamara.Ginige@acu.edu.au
5
Macquarie University, Australia, deborah.richards@mq.edu.au
6
University of Salerno, Italy, {pdigiovanni, msebillo, gvitiello, tortora}@unisa.it
7
University of California, Irvine, USA, jain@ics.uci.edu
Abstract— Crop production problems are common in Sri
Lanka which severely effect rural farmers, agriculture sector and
the country’s economy as a whole. A deeper analysis revealed
that the root cause was farmers and other stakeholders in the
domain not receiving right information at the right time in the
right format. Inspired by the rapid growth of mobile phone usage
among farmers a mobile-based solution is sought to overcome
this information gap. Farmers needed published information
(quasi static) about crops, pests, diseases, land preparation,
growing and harvesting methods and real-time situational
information (dynamic) such as current crop production and
market prices. This situational information is also needed by
agriculture department, agro-chemical companies, buyers and
various government agencies to ensure food security through
effective supply chain planning whilst minimising waste. We
developed a notion of context specific actionable information
which enables user to act with least amount of further processing.
User centered agriculture ontology was developed to convert
published quasi static information to actionable information. We
adopted empowerment theory to create empowerment-oriented
farming processes to motivate farmers to act on this information
and aggregated the transaction data to generate situational
information. This created a holistic information flow model for
agriculture domain similar to energy flow in biological
ecosystems. Consequently, the initial Mobile-based Information
System evolved into a Digital Knowledge Ecosystem that can
predict current production situation in near real enabling
government agencies to dynamically adjust the incentives offered
to farmers for growing different types of crops to achieve
sustainable agriculture production through crop diversification.
Keywords—Mobile-based Information System for Farmers;
Digital Knowledge Ecosystem for Agriculture; Sustainable
Agriculture Production
I.
I
NTRODUCTION
A major agricultural problem in Sri Lanka is
overproduction of vegetables due to many farmers growing
the same crop in the same area without having an awareness of
what others’ are growing [1, 2]. This also leads to
underproduction of some other crops. The net result is waste
due to over production, wide price fluctuations at the farm
gate as well as in retail outlets and farmers getting trapped in a
poverty cycle. Thus in August 2011, an international
collaborative research team consisting of researchers from
seven universities in four continents embarked on a project to
explore a possible solution for farmers in Sri Lanka to address
this problem associated with crop cultivation [3].
A deeper analysis revealed that the root cause for the
problem was farmers as well as other stakeholders in the
domain not receiving right information at the right time in the
right format. This led us to explore possible mechanism for
effective delivery of required information. In Sri Lanka there
has been a rapid growth in mobile phone usage in the past few
years. According to Telecommunications Regulatory
Commission of Sri Lanka, the number of Cellular Mobile
Subscribers at the end of December 2010 (time when this
project was initiated) was around 80.95% of the total
population which has increased to around 116.7% of the total
population by December 2015 [4]. Even the farmers in rural
areas, irrespective of their native language, tend to carry a
mobile phone to the farm, and use it in their day to day
activities. Further Sri Lanka's population has a literacy rate of
92%, one of the highest literacy rates in Asia [5]. Inspired by
this wide spread use of mobile phones and the user
empowerment provided by Social Networking applications
such as Facebook and Twitter [6], we decided to explore a
possible mobile-based solution to address this problem.
Social networks provide us with tools to share our
experiences using text and a variety of other mediums [7]. The
ways users can participate in these networks have significantly
changed as a result of the second generation of Web
applications known as Web 2.0. These applications allow
users not only to view content but also to add content. Thus
users who were previously passive information consumers
have been empowered to become information producers
facilitating effective two way flow of information [8]. Jain and
Sonnen have shown that it is now possible to aggregate
2016 IEEE International Conference on Data Science and Advanced Analytics
978-1-5090-5206-6/16 $31.00 © 2016 IEEE
DOI 10.1109/DSAA.2016.82
602

information from various sensors on a mobile phone, other
published data sources and micro blogs such as twitter, to
detect evolving situations and make this information available
to the users in real time [9]. These networks extend the
information sharing concept in Social Networks and are called
Social Life Networks (SLN).
We first conceptualised a possible solution to the
overproduction problem based on SLN concept where farmers
use a Mobile Based Information System (MBIS) to report the
extent of their crop cultivation. This information is then
aggregated based on location, time and crop type to derive
current production levels for different crops in real time. This
aggregated information is made available to farmers to make
an informed selection of crops to grow in coming farming
season.
The next section presents the important aspects of the
physical solution that evolved from this conceptualisation and
the insights gained. Section III explains how the evolved
solution was reconceptualised as a Digital Knowledge
Ecosystem followed by the growth it experienced since its
deployment in April 2015. Then we present how it became
possible to develop a conceptual solution to manage the
agriculture production as a closed loop system for the whole
country to achieve sustainability, how it compares with similar
systems followed by the conclusions.
II. M
OBILE
-
BASED
I
NFORMATION
S
YSTEM FOR
S
RI
L
ANKAN
F
ARMERS
Design Science Research (DSR) methodology was
selected for this project as it is well suited to create an artefact
as a solution to a research problem [10]. In this project the
main artefact to be developed was a mobile-based information
system for farmers. In this process, many supporting artefacts
were also created in the form of architectures, models and
theories. DSR consists of 3 cycles: Relevance, Design and
Rigour. Researchers can move around these cycles creating
and validating the artefacts that get produced. The research
team went through many iterations of these cycles to develop
the system [11, 12].
Figure 1 shows the order in which these cycles took place
to create the mobile artefact. In the figure, R (1 - 5) represent
various Relevance Cycles. In Relevance Cycles, researchers
spend time in the environment in which the artefact get used
(in this cases with farmers and other agriculture domain
stakeholders) to understand the deeper needs and evaluate the
suitability of both the ideas and artefacts that were developed
to solve the problem. Rigour Cycles are not shown in the
diagram. In Rigour Cycles, researchers review the related
scientific literature and identify relevant theories, models and
architectures that can be useful to solve the problem. Once a
solution is developed, researchers share the newly gained
knowledge and insights in terms of papers and seminars. In
Design Cycles D (1 - 7), researchers perform a heuristic search
process guided by theories, models and architectures
uncovered during the corresponding Rigour Cycle to find a
solution to a problem identified in a Relevance Cycle. If the
outcome of the design process is a functional artefact it is
checked to see whether the artefact is functionally correct.
The need and order of these cycles evolved based on the
findings and/or required validations resulting from previous
cycles. The project started with a Relevance Cycle R1 to get a
deeper understanding of the domain and issues leading to
various crop production problems. To manage the complexity
we initially limited our investigation to overproduction
scenario.
Fig. 1. The order of DSR Cycles that evolved to develop the MBIS
Below is a description of important aspects of this iterative
development process.
A. Application Domain Analysis – R1
The initial overproduction scenario led the team to
investigate what information is available for farmers to make
decisions, especially to select a crop to grow in the coming
season. We surveyed farmers to better understand the factors
that influence their selection of crop(s) to grow as well as their
familiarity with technology. Results showed that 92% of the
farmers owned a mobile phone and nearly 50% of them used it
to obtain market information. Over 75% farmers stated that
there was no proper mechanism for them to get the required
agricultural information on time. It was found that due to lack
of access to current information; farmers relied on past
information such as market price of a crop in the previous
season and their own experiences [13].
We found that information needs of farmers can be divided
into two broad categories: Quasi Static and Dynamic or
Situational. Quasi Static information includes crop varieties,
how to grow different varieties, types of fertilizer to be used,
how to manage pests, diseases etc. These types of information
evolve slowly with time; hence the term quasi static. Farmers
also wanted information such as local weather, market prices,
current production of various crops, current demand, nearest
location from where they can purchase seeds, fertilizer and
corresponding price information which changes rapidly; hence
the term Dynamic information or Situational Knowledge [12,
13].
B. Visualising the initial Conceptual Solution – D1
Since the aim of the research project was to develop an
artefact, the first Design Cycle was to visualise a possible
603

physical realisation of the conceptual solution that we
formulated earlier. We used the scenario transformation
approach of Rosson and Carrol to design the first set of
interfaces [14]. In this approach, we created a set of typical
scenarios and personas of actors based on data gathered from
the surveys [15]. Next we investigated how information
deficiencies in these scenarios can be mitigated by providing
missing information compared to information needs identified
through surveys and literature, and created a set of
transformed scenarios. Based on original scenarios and
personas we also identified the usability requirements. We
used transformed scenarios and usability requirements to
develop the first set of user interfaces shown in Fig 2. The first
screen represents a crop catalogue. Icons were used to
describe crops and a coloured background to indicate the
approximate quantity of each crop already in production. The
colour scale ranges from white indicating zero production to
red indicating intensive production.
Fig. 2. The crop catalogue and product select interface.
After selecting a crop, users can navigate to the second
interface to obtain a more detailed description of the product.
We provided check box items to allow users to insert
information about the quantity of crop(s) that they wanted to
cultivate to eliminate typing errors. This information could
then be aggregated to derive the current production levels in
real time and used to decide the colour codes for crops in the
crop catalogue. For example the most cultivated crop(s) can be
shown in red colour and the least cultivated/selected crop(s)
can be shown in white colour. This design gave the whole
research team a good idea about how information can be
visualised and user input can be captured. We also realised
that it is very effective to use colour to communicate the
current production levels to the farmers.
This visualisation of a potential implementation of the
conceptual solution led us to formulate the following sub
research questions.
• What information farmers need, when do they need it
and what is an effective format to present it?
• How can we motivate framers to act on this
information?
• How can we obtain or generate information that
farmers need?
C. Information Flow Model
The physical visualisation of the conceptual solution
assisted us to identify a possible information flow model to
meet farmer information needs. Once a farmer registers with
the system by providing his/her details the first step is to
provide information to assist registered farmer to select a crop.
Next the planned extent needs to be captured and using this
information total production for the region can be predicted
which is to be provided back to farmers.
Most of quasi static information necessary for farmers to
select a crop is available from various Government agencies in
the form of booklets, TV and seminar programs. However, a
farmer needed to perform a considerable amount of cognitive
processing on this information to search and find the relevant
information to select a suitable crop for his or her farm. This
led us to formulate the concept of “actionable information”
that is at a level of granularity that a person can act with least
amount of further processing [16]. This information needs to
be provided in context. In the case of a farmer querying what
crops will grow in my farm, we need to model the farm
context with parameters/features such as rainfall, temperature,
elevation and soil condition which would vary based upon the
location of the farm. This enables to provide a list of crops that
will grow under these conditions. Further if we know a farmer
has specific preferences such as cash crops and vegetables the
list can be further narrowed down based on farmer preference.
Thus we identified the need to provide context specific
actionable information to farmers. This led us to investigate
two other issues namely what is a suitable context model and
how to obtain parameter values for variables in the model that
are specific to a user.
D. Ontological Knowledge base Development – D2
As identified above farmers want to know “what crops will
grow in my farm”. “My farm” can be characterised by climatic
conditions for the planned season, type of soil and water
availability at the location. This forms the context for the
query. The Agriculture Department: a major stakeholder in the
domain has published how to grow various crops, required
climatic conditions, suitable soil type, watering methods,
fertilizer and pesticides to be used as documents on their
website [17] and booklets. Since this knowledge is scattered
and in document form it is not possible to query it in context.
To ask the question “what crops will grow in my farm”, first it
is necessary to expand “my farm” with corresponding values
stored in the context model. Next it is necessary to
disaggregate the information in the website or documents into
a suitable granular representation so that one can query this
information in context. For this an ontological knowledge base
was developed [18-21].
First a set of frequently asked questions by
farmers/stakeholders were identified. Some of these questions
are “what crops will grow in my farm”, “What are the best
varieties”, “when is best time to apply fertilizer”, “how much
of a particular cop has been planted”. Next as shown in Table
1 the context information is automatically added to the query
604

based on the location of the farm. To identify some of the
parameters in the context model the MBIS used the GPS
capability in a Smartphone. The system captures the geo-
coordinates and maps the location of the farm onto an agro-
ecological zone map. Each agro-ecological zone has specific
climatic and weather conditions. These values specify the farm
context that needs to be used to query the ontological
knowledge base [22, 23]. From these, generalised
contextualised queries were generated and expressed this
using first-order logic (FOL) (refer Table 1).
For the ontology development, the Grüninger and Fox’s
methodology was selected. Being a formal ontology it is
structurally and functionally rich to describe the domain
knowledge in context. For the implementation, Protégé as the
ontology development environment and Web Ontology
Language (OWL) as the ontology language were selected.
Protégé OWL plugin combination is good as a tool for
ontology creation because of its scalability and extensibility.
Protégé also has powerful frames and its user interface
provides an easy to use environment. Since Description Logics
(DL) is a fully decidable fragment of FOL and also reduces
the complexity when compared with FOL, the DL based
approach (OWL 2-DL) is selected to implement the ontology.
In this implementation, decidability is very important as we
need to retrieve agricultural information and knowledge in
user’s context.
TABLE 1 Deriving Actionable Information
Farmers’
Information
Needs
Farmers’
Information Needs
in Context
Generalising
Contextualised
Information
Query in First
Order Logic
(FOL)
What are the
suitable
crops to
grow?
Suitable crops based
on the Environment:
• Which crops are
suitable to grow in
the ‘Dambulla’
area?
• What are the
suitable vegetable
crops for
‘UpCountry’,
applicable to the
‘Well-drained
Loamy’ soil, and
average rainfall >
2000 mm?
Suitable crops based
on Preferences of
Farmers:
• What Brinjal’s
varieties can resist
the ‘Bacterial Wilt’
disease?
• Which crops
are suitable to
grow in
specified
Location?
• What are the
suitable Types
of Crops for
specified
Location,
applicable to
the specified
Soil
types/character
istics, and other
Conditions?
• What Crop’s
varieties can
resist the
specified
Disease?
(∃x)(Crop(x)) ∧
RegionalArea(Da
mbulla) ∧
grows(x,
Dambulla);
(∃x)(Vegetable(x)
)∧
SoilType(Loamy)
∧ SoilDrainage(
Well_drained)∧ h
asSoilFactor
(x,Loamy) ∧
hasSoilFactor(x,
Well_drained)∧
(∃y Integer(y)
∧ hasMinRainfall
(x,y)∧ (2000 ≤
y));
For online knowledge base creation, Resource Description
Framework (RDF) was used. A Semantic Web toolkit: ARC2
(appmosphere RDF classes as a SPARQL endpoint) is used to
manage the RDF data. The online knowledge base with a
SPARQL endpoint was created to share and reuse the domain
knowledge that can be queried based on user context.
(http://webe2.scem.uws.edu.au/oms/searchInformation.php)
E. Motivating Farmers to Act – D3
Next we needed to address how to motivate the
stakeholders to act on the information in order to generate
transaction data. It was noted that rapid growth of Social
Networks can be attributed to user empowerment as a result of
symmetric information flow and the associated
communication and collaboration patterns [24]. This led us to
carry out a detailed study of psychological empowerment to
understand the various components of psychological
empowerment, key drivers and influences of empowerment
and their relationships to each other [25, 26]. Empowerment at
the individual level of analysis can be referred to as
psychological empowerment (PE). Zimmerman defines three
qualities of psychological empowerment (PE) [26].
Intrapersonal Component of PE refers to how people think
about themselves and includes domain specific perceived
control and self-efficacy, motivation to control and perceived
competence. Interactional Component of PE refers to the
understanding that people have about their community and
related socio-political issues. Behavioural component of PE
refers to actions taken to directly influence outcomes. These
three components of PE merge to form a picture of a person
who believes that he or she has the capability to influence a
given context (Intrapersonal Component), understands how
the system works in that context (Interactional Component)
and engages in behaviours to exert control in the context
(Behavioural component).
Some key drivers and influencers of psychological
empowerment are decision making, choices, autonomy,
engagement, communication and meaningful personal goals.
In the study of human motivation, meaningful goals have been
identified as the key contributing factor to the long-term levels
of well-being [27-30]. A person can gain a greater sense of
control in his/her life is a consequence of exercising greater
choices [31]. There is a growing evidence to support the
importance of individuals' beliefs concerning their capabilities
to exercise control over important aspects of their lives [32-
36]. The levels of intrapersonal component of PE exhibits
different behaviour patterns of individuals such as how they
respond to choices, make decisions, engage in their
community and society. Empowerment is an iterative process
in which a person takes action to achieve a personally
meaningful goal and then reflects on the impacts of that
action. The levels of the three qualities of psychological
empowerment are continually changing. This we can call the
empowerment flow as shown in Fig. 5.
For example, if farmers share the extent of the crops that
they have planted (exhibit behaviour), by aggregating this
information the system can predict current production level.
This information can be made available to other farmers to
make an informed decision when selecting a crop(s) to grow.
This can result in diversification of crops being grown in order
to reduce the overproduction problem. Thus farmers have
obtained a beneficial outcome as a result of the action they
took (positive feedback from the environment). This could
then motivate farmers to keep sharing the information.
Based on above insight we created tools to assist farmers
to act on the information to achieve their goals. Primary goal
605

of the majority of farmers is to achieve a good revenue. Thus
we initially created a “Profit Calculator” to compute
agricultural expenses and to keep track of the history of such
expenses. This evolved into a comprehensive “Cultivation
Planning” use case as describe in sub section H.
F. Generating Dynamic Content – D4
The next step was to investigate how actionable
information relating to the current situation required by
stakeholders can be dynamically generated from transaction
data captured through action taking. For this we identified
information needs of major stakeholders in the agriculture
domain and the information they would generate by acting on
this information. Through this mapping the research team
discovered information needs of various stakeholders
(consumption) can be derived by information generated by
some other stakeholders as part of their activities (production)
[37]. Often there is no direct mapping between information
production and consumption. This mapping happens through
two operators: Aggregation and Disaggregation.
The following list describes this mapping.
• Farmers provide planned extent to be cultivated as part of
calculating the cost of production using “Cultivation
Planning” function and they get immediate benefit by
being able to find the cost. This information is aggregated
and provided back as “total expected production” to
farmers and agriculture department officers. Providing this
information to farmers who are in the process of selecting
a crop to grow can assist crop diversification. Government
can use this information for issuing import permits for
different food types to maintain food security in the
country.
• Once farmers select various fertilizer, pesticides, seeds and
other chemicals again as part of calculating the cost of
production that information can be aggregated to predict
the demand for these items in the coming months for
various geographical regions. Agro-chemical companies
can use this information to plan their supply chains. In
return they have started providing their price list through
the system. We plan to incorporate an online ordering
facility in the future.
• Having seen the potential of the system; especially for
creating an online market place a couple of banks have
approached the research team to explore how they can
provide micro finance through the system in case farmers
have a cash flow difficulty.
• Farmers having seen the possibility of an online market
place during field trials also requested an option for them
to sell seedlings, fertilizer and particularly organic
fertilizer which some of them are producing in commercial
quantities. In response to this request “My Offerings” use
case was designed for farmers to enter what they would
like to sell to other farmers. When farmers use expense
calculator, if the required seedlings, fertilizers etc. are
offered by another farmer living within a pre-defined
radius this information and contact details of the farmer
selling` the items is displayed.
G. First Functional Prototype and Field Trial D5 and R2
We created the first functional prototype by combing the
artefacts designed in various design cycles. We organised the
functionality into 3 modules; “Login”, “Crop Selection” and
“Profit Calculator” particularly considering the crop selection
phase in farming life cycle and thus the computation of the
expenses associated with each such selection.
This prototype was field trialled in December 2012 in a
main vegetable producing region in Sri Lanka with 32
farmers. We found that 56% of the farmers were attracted to
the existing production levels presented using the colour
coding scheme as described before. Around 47% of the
farmers found the information provided with respect to crop
types and their different varieties very useful, but wanted more
details about crop characteristics. Some farmers mentioned the
importance of showing more information such as the price
sold and the issues faced in the previous seasons thus
highlighting the need to maintain historical data.
It was also observed only about 50% of the farmers have
systematically calculated the intended cost of the planned
cultivation for the coming season. All the farmers had an
approximate idea about what their net revenue might be.
Around 81% of the farmers mentioned that having a better
understanding of their expenses in the various stages of the
crop cycle and an awareness of different suppliers may help
them to better manage their expenses.
H. Revised Design and 2nd Field Trial – D6 and R3
We analysed the findings in detail and refined the
application accordingly [12, 38, 39]. The information provided
during the crop selection stage is enhanced by extending the
ontology to include more details about the varieties and some
common diseases. The profit calculator evolved into a
comprehensive “Cultivation Planning Use case” with a built in
expense calculator and expense history module. Once a farmer
selects a crop we queried the crop ontological knowledge base
to obtain the required fertiliser and pesticides and the
recommended quantities. We further added a Farm Supplies
database into the system to show the potential suppliers and
their prices to farmers. In order to provide the additional crop
information as well as associated fertiliser and pesticide
information in the context of farm environment, the crop
ontology was significantly expanded [19].
After making the above changes the prototype system was
field tested at 3 locations in Sri Lanka in November 2013
involving 50 farmers. Farmers were very happy with the
extended functionality. We observed some usability issues
relating to gestures farmers are using to interact with the
Smartphone.
I. Deployed System – D7 and R4
Based on the 2nd field trial findings (R3) we further
refined the system. In order for farmers to be able to sell their
seeds, fertiliser and to hire out labour and equipment we added
a new use case “My Offerings” to the existing use cases of
“Crop Selection” and “Expense Calculator”. We hope this
would facilitate the creation of a local market to further
empower farmers.
606

User interfaces for this crop section use case is shown in
Fig. 3. The user has to first register a farm. A user can have
many farms. When registering a farm the system captures the
farm’s geo-coordinates. Then using a backend GIS system that
is aware of agro-ecological zones in Sri Lanka the system
finds the corresponding agro-ecological zone. The system then
queries the ontological knowledge base to find the crops and
varieties that grow in that agro-ecological zone and present the
list to the user to select which crops to grow.
User interfaces for the Expense Calculator use case is
shown in Fig. 4. The user needs to provide the planned extent
of the selected crops(s) to enter the Expense Calculator. In the
Expense Calculator the farmer first selects the farming stage.
Then the farmer is provided information on fertilizer,
pesticide, seeds and other chemical requirements obtained
from the ontological knowledge base and the required quantity
calculated based on planned extent earlier provided. Agro-
chemical companies can enter the products that they have to
offer to the farmers as well as the retail price. At this stage
farmers are presented with a list of suppliers to select from.
Once they select a supplier based on retail price information in
the supplier database the cost of growing the selected crop
variety for the selected extent is calculated. Based on this
information farmers can select what crops to grow.
Login Screen Main Menu Select a Farm Add a new farm using
address
Detail information of a
selected variety
Varity Information Crops that will grow in
“My Farm”
Add a new farm using a
map
Fig. 3. User Interfaces for Crop Selection Use Case – “What will grow in my farm?”
Specify planned
extent
Select Stage to calculate
expense
Select the expanse type
to be Calculated
Select Supplier Summary of Costs
Fig. 4. User Interfaces for Expense Calculator Use Case – “How much it will cost?”
607

III. R
ECONCEPTUALISING AS A
D
IGITAL
K
NOWLEDGE
E
COSYSTEM
The initial aim of developing the MBIS was to minimise
the overproduction problem by providing necessary
information to make informed decisions. As a result of using a
user-centred design concept the system evolved into providing
information to assist many other activities in a crop cycle [11,
15, 40, 41]. When attempting to find an efficient way to
source the required information a complex information flow
model for the agriculture domain evolved connecting all major
stakeholders [13, 37]. At this stage the information flow model
of the MBIS started to resemble energy and nutrient flow
model of biological ecosystems.
The concept of biological ecosystems was first proposed
by A.G Tansely in the 1930s [42] and has been studied by
many since then. A biological ecosystem is a community of
living organisms (plants, animals and microbes) in
conjunction with the nonliving components of their
environment (things like air, water and mineral soil),
interacting as a system linked together through nutrient cycles
and energy flow [43].
Similarly the MBIS that evolved for Sri Lankan farmers
consisted of major stakeholders of the agriculture domain
which formed the community, linked together through
information flows with the help of various ICT technologies.
In biological ecosystems energy gets aggregated when animals
feed on plants and one another. Decomposers such as worms
and microbes decompose the dead animals and biomass
releasing this energy and converting matter to a form that can
again be absorbed mainly by plants creating a continuous flow
pattern [43]. Similarly to create an information flow pattern
among the main stakeholders of the agriculture domain we had
to introduce information aggregation and disaggregation
modules into our system. Like decomposers convert matter to
a form that can again be absorbed, MBIS breaks down large
chunks of information to actionable information. Thus the
architecture that evolved for MBIS shown in Fig 5 can be
considered as a form of an ecosystem. Thus we
reconceptualised the system that evolved as a Digital
Knowledge Ecosystem (DKES).
Fig. 5. Schematic Diagram of the Information and Empowerment flow patterns – Information flow is shown in black and empowerment flow is shown in red
IV. T
HE
G
ROWTH OF
DKES
The behaviour of an ecosystem is of two parts: dynamic
(ecology) and evolutionary. The Dynamic behaviour
represents how stakeholders (agents) are interacting with the
system. Agents based on available information perform tasks,
in the process making decisions, to achieve their individual
goals. The aggregated action produces the overall domain
outcomes (for example overall agricultural yield for the
season). The system should provide actionable information
and also capture the decisions they make when performing the
actions to create the information flow model. Ecosystems start
small and then grow. Evolutionary behaviour is a specific
characteristic of an ecosystem that helps to grow the
population as well as to attract new communities (example –
Stakeholder groups such as farmers, agriculture extension
officers, agro-chemical companies etc).
Fig. 6. Growth of MBIS Communities since its launch in April 2015
608

Since the launch of the MBIS in April 2015 we started to
observe a growth pattern as shown in Fig. 6. The initial growth
was coming from different communities joining the
ecosystem.
V. C
LOSED
L
OOP
M
ANAGEMENT OF
A
GRICULTURE
P
RODUCTION
As seen in Sri Lanka and in other parts of the developing
world farmers are vulnerable to many factors: weather and
mismatch between supply and demand being dominant among
these. Unlike any other industry, in agriculture the farmers
face a unique problem – the ‘production lag’. This means that
between the time of production and the time of consumption,
there is considerable time lag. Thus any mismatch between
demand and supply is difficult to correct and would create an
adverse price movement. As a result of these fluctuations, both
the farmer and the consumer may suffer. The root of the
problem is the poor coordination between the market and the
farmers. Thousands of farmers independently make choices
about the crop types and production quantities. These
individual decisions when taken collectively may not exactly
be what the market demands. In other words, farmers rarely
make optimal decisions that lead to perfect balance between
total production and market demand. Therefore, there needs to
be a third party who coordinates the activities of the farmers
and the market. In Sri Lanka this role is played by the
government by providing incentives to direct the production
towards certain crops that otherwise will be in short supply.
Typically these incentives were in the form of fertilizer
subsidies [44] or guaranteed purchase prices. These incentives
have worked to increase the harvest of targeted crops. The
problem is if the incentives lead to an overproduction situation
there is no mechanism to detect this early in order to take
corrective action. How much has been produced is accurately
known only when the harvest comes to the market. Thus the
current situation can be conceptualised as an open loop control
system. When there are many unpredictable variables; in this
case weather and availability of finance being two major once,
an open loop control system will not give the desired market
stability.
The Digital Knowledge Ecosystem that evolved for Sri
Lankan agriculture domain provides essential actionable
information to farmers at various stages of the farming life
cycle and captures production related information in real time.
Using suitable predicative algorithms it is now possible to
compute in near real time how much of each crop type is
being planted in different parts of the country as well as the
future needs for fertilizer and pesticide based on
recommended application levels again for various parts of the
country. The Government can use the predicted production
levels to dynamically adjust the available incentives for
different crops to archive crop diversification and a better
balance between supply and demand. Agro-chemical
companies can use the predicted future needs for various
inputs to optimally plan their supply chains. Major buyers can
use this system to have stable supply of agriculture products at
an agreed quality and price. These will reduce waste
benefiting all the stakeholders. Stable prices and the resulting
improvements to livelihoods of farmers will encourage more
people to take up agriculture. All these will contribute to
archive sustainability. This conceptual model is shown in Fig.
7.
Based on this conceptual model we submitted a proposal to
develop this system to the Sri Lanka Government in
September 2015. A National Project “Govi Nena - Agriculture
Intelligence” was announced in the National Budget speech in
November 2015 [45]. This will be the first time such a closed
loop system will be developed at national level in a country to
achieve food security for the country and stable and
sustainable prices for farmers and consumers.
Fig. 7. Conceptual Closed Loop Agriculture Production Management System
609

VI. C
OMPARISON WITH SIMILAR SYSTEMS
In literature, there are many mobile-based agriculture
information systems to assist farmers to carry out their
agriculture value chain activities. These can be categorised
into extension and knowledge systems, market information
systems, procurement and traceability systems or inspection
and certification systems [46]. AvaajOtalo (voice stoop)
[47], eSagu agro-advisory solution [48], mKrishi [49] are
few mobile based extension and knowledge systems in
India. The KACE [50], Esoko [51] and MFarm [52] provide
market information to link farmers and buyers. Existing
procurement and traceability systems were designed to
reduce transportation cost which is the main contributor to
the transactional cost spend by rural farmers in developing
countries [46]. None of these provide a holistic information
flow model connecting all major stakeholders in the
agriculture domain.
VII. C
ONCLUSION
This paper presents an approach to achieving a
sustainable agriculture production for a whole country using
a Digital Knowledge Ecosystem to enhance the flow of
information in the agriculture domain. The distinguishing
feature in this approach is that it is based on a closed loop
management approach rather than a traditional open loop
approaches. In open loop approaches required production
extents were pre decided by relevant government agencies
and farmers are encouraged to grow these amounts by
providing various incentives or subsidy schemes. Any
deviations that can happen to the forecast yield due to
farmers not adhering to the recommended crop varieties and
extents, unexpected weather conditions, issues with
fertilizer, pest and disease problems during the interim
period between planning and harvesting often goes
undetected until harvests come to the market. By this time it
is too late to take any corrective measures.
This Digital Knowledge Ecosystem evolved as a
solution to the overproduction problem faced by farmers in
Sri Lanka. Our research showed overproduction is a
symptom of a much deeper problem; farmers as well as
other stakeholders not getting necessary information at the
right time in the right format to make informed decisions.
The stakeholders needed published information (quasi
static) about crops, pests and diseases, land preparation,
growing and harvesting methods. They also needed real-
time situational information (dynamic) such as current crop
production, market prices etc. In this project we devised a
method based on user centred Ontology to convert
published quasi-static information into actionable
information. We adopted empowerment theory to create
empowerment-oriented farming processes to motivate
farmers to act on this information. By aggregating the
transaction data resulting from farmer actions we generate
situational information required by all stakeholders to create
a complete information flow model for the domain
mimicking energy flow in biological ecosystems.
The ability to obtain aggregated production levels in
near real-time opened up the possibility for relevant
government agencies to monitor the crop production in real-
time, adjust the incentives dynamically to achieve planned
yields and communicate the available incentives through the
system back to farmers at the time of selecting a crop to
grow. Thus it became possible to manage the overall
agriculture production of a country based on a closed loop
system.
The Sri Lankan Government has embraced this
conceptual possibility and has announced a National Project
“Govi Nena - Agriculture Intelligence” to implement this
closed loop monitoring system for achieving sustainable
agriculture production.
A
CKNOWLEDGMENT
We would like to acknowledge the financial assistance
we received from the National Science Foundation of Sri
Lanka, HRD Program of the HETC project of the Ministry
of Higher Education, Sri Lanka various monetary as well as
in kind support from participating Universities in this
International Collaborative Research program. Also the
support given to us by the farmers and agriculture extension
officers in Sri Lanka is greatly appreciated.
R
EFERENCES
[1] S. Hettiarachchi, "N'Eliya carrot farmers in the dumps: Bumper
harvest, but prices low," in The Sunday Times ed. Sri Lanka, 2012.
[2] S. Hettiarachchi, "Leeks Cultivators Desperate as Price Drops to
Record Low," in Sunday Times, ed. Sri Lanka, 2011.
[3] A. Ginige. (2011, 25/03/2012). Social Life Networks for the Middle of
the Pyramid. Available:
http://www.sln4mop.org//index.php/sln/articles/index/1/3
[4] TRC. (2015). Telecommunications Regulatory Commission of Sri
Lanka - Statistics. Available: http://www.trc.gov.lk/2014-05-13-03-
56-46/statistics.html
[5] Wikipedia. (2016). Education in Sri Lanka. Available:
https://en.wikipedia.org/wiki/Education_in_Sri_Lanka
[6] M. D. Fernando, A. Ginige, and A. Hol, "Impact of Social
Computing on Business Outcomes," presented at the 13th
International Conference on Web Based Communities and Social
Media, Funchal, Madeira, Portugal, 2016.
[7] M. Giles. (2010) A special report on social networking: A world of
connections. The Economist. S1-S20. Available:
http://www.economist.com/node/15351002
[8] A. Ginige and M. D. Fernando, "Towards a generic model for social
computing and emergent characteristics," in 2015 2nd Asia-Pacific
World Congress on Computer Science and Engineering (APWC on
CSE), 2015, pp. 1-10.
[9] R. Jain and D. Sonnen, "Social Life Networks," IT Professional vol.
13, pp. 8 - 11, 2011.
[10] A. Hevner and S. Chatterjee, "Design science research in information
systems," in Design Reasearch in information systems, ed: Springer
science+Business Media, 2010, pp. 9-21.
[11] L. D. Silva, T. Ginige, P. D. Giovanni, M. Mathai, J. Goonetillake, G.
Wikramanayake, et al., "Interplay of Requirements Engineering and
Human Computer Interaction approaches in the Evolution of a
Mobile Agriculture Information System," in Usability and
Accessibility focused Requirements Engineering: Bridging the Gap
between Requirements Engineering and Human-Computer
Interaction. vol. 9312, 2016, A. Ebert, S. R. Humayoun, N. Seyff, A.
Perini, and S. D. J. Perini, Eds., ed: Springer 2016.
[12] L. N. C. De Silva, J. S. Goonetillake, G. N. Wikramanayake, and A.
Ginige, "Farmer Response towards the Initial Agriculture
Information Dissemination Mobile Prototype," in Computational
Science and Its Applications – ICCSA 2013. vol. 7971, B. Murgante,
610

S. Misra, M. Carlini, C. Torre, H.-Q. Nguyen, D. Taniar, et al., Eds.,
ed: Springer Berlin Heidelberg, 2013, pp. 264-278.
[13] L. N. C. De Silva, J. S. Goonetillake, G. N. Wikramanayake, and A.
Ginige, "Towards using ICT to Enhance Flow of Information to aid
Farmer Sustainability in Sri Lanka," presented at the 23rd
Australasian Conference on Information Systems Geelong, Victoria,
Australia, 2012.
[14] M. B. Rosson and J. M. Carroll, Usability engineering: scenario-
based development of human-computer interaction: Morgan
Kaufmann Publisher, 2002.
[15] P. D. Giovanni, M. Romano, M. Sebillo, G. Tortora, G. Vitiello, L.
D. Silva, et al., "User Centered Scenario based Approach for
Developing Mobile Interfaces for Social Life Networks," presented at
the 34th International Conference on Software Engineering (ICSE
2012) - UsARE Workshops, Zurich, Switzerland, 2012.
[16] A. Ginige, "Digital Knowledge Ecosystems: Empowering Users
through Context Specific Actionable Information," presented at the
9th International Conference on ICT, Society and Human Beings
(ICT 2016), Madeira, Portugal, 2016.
[17] Agriculture. (2011, 28/11/2011). Department of Agriculture
Govenment of Sri Lanka. . Available: http://www.agridept.gov.lk/.
[18] A. I. Walisadeera, A. Ginige, and G. N. Wikramanayake, "User
centered ontology for Sri Lankan farmers," Ecological Informatics,
vol. 26, pp. 140-150, 2014.
[19] A. I. Walisadeera, A. Ginige, and G. N. Wikramanayake,
"Conceptualizing Crop Life Cycle Events to Create a User Centered
Ontology for Farmers," in International Conference on
Computational Science and its Applications (ICCSA 2014), Poroto,
Portugal, 2014, pp. 791–806.
[20] A. I. Walisadeera, G. N. Wikramanayake, and A. Ginige, "An
Ontological Approach to Meet Information Needs of Farmers in Sri
Lanka," presented at the 1st International Workshop on Agricultural
and Environmental Information and Decision Support Systems
(AEIDSS 2013), Ho Chi Minh City, Vietnam., 2013.
[21] A. I. Walisadeera, G. N. Wikramanayake, and A. Ginige, "Designing
a Farmer Centred Ontology for Social Life Network," presented at
the 2nd International Conference on Data Management Technologies
and Applications (DATA 2013), Reykjavík, Iceland, 2013.
[22] M. Mathai and A. Ginige, "Task Oriented Context Models for Social
Life Networks," in Software Technologies, ed: Springer Berlin
Heidelberg, 2014, pp. 306-321.
[23] M. Mathai and A. Ginige, "Context based Contenet Aggregation for
Social Life Networks," presented at the 8th International Conference
on Software Paradigm Trends (ICSOFT-PT 2013), Reykjavik,
Iceland, 2013.
[24] R. Hoegg, R. Martignoni, M. Meckel, and K. Stanoevska-Slabeva,
"Overview of business models for Web 2.0 communities,"
Proceedings of GeNeMe, vol. 2006, pp. 23-37, 2006.
[25] M. A. Zimmerman, "Citizen participation, perceived control and
psycological empowerment " American Journal of Community
Psychology, vol. 16, pp. 725-750, 1988.
[26] M. A. Zimmerman, "Psycological Empowerment: Issues and
Illustrations," American Journal of Community Psychology, vol. 23,
pp. 581-600, 1995.
[27] J. T. Austin and J. B. Vancouver, "Goal constructs in psychology:
Structure, process and content," Psychological Bulletin, vol. 120, pp.
338-375, 1996.
[28] P. Karoly, "A goal systems - self-regulatory perspective on
personality, psychopathology and change," Review of General
Psychology, vol. 3, p. 264, 1999.
[29] R. Alsop and N. Heinsohn, "Measuring Empowerment in Practice:
Structuring Analysis and Framing Indicators," World BankFebruary
2005.
[30] E. L. Deci and R. M. Ryan, "The "what" and "why" of goal pursuits:
Human needs and the self-determination of behaviour," Journal of
Psychological Inquiry, vol. 11, pp. 227-268, 2000.
[31] E. J. Lawler, "Affective attachments to nested groups: A choice-
process theory," American Sociological Review, pp. 327-339, 1992.
[32] A. Bandura, "Self Efficacy Mechanism in Human Agency,"
American Psychologist, vol. 37, pp. 122-147, 1982.
[33] L. Corno and E. B. Mandinach, "The role of cognitive engagement in
classroom learning and motivation," Educational Psychologist, vol.
18, pp. 88-108, 1983.
[34] C. S. Dweck and E. L. Leggett, "A Social-Cognitive Approach to
Motivation and Personality " Psychological Review vol. 95, pp. 256-
273, 1988.
[35] D. J. Stipek and J. R. Weisz, "Perceived Personal Control and
Academic Achievement," Review of Educational Research, vol. 51,
pp. 101-137, 1981.
[36] D. H. Schunk, "Peer Models and Children’s Behavioral Change,"
Review of Educational Research, vol. 57, pp. 149 - 174, 1987.
[37] L. De Silva, J. Goonetillake, G. Wikramanayake, and A. Ginige,
"Towards an Agriculture Information Ecosystem," in 25th
Australasian Conference on Information Systems,, 2014.
[38] T. Ginige. and D. Richards., "A model for Enhancing Empowerment
in Farming using Mobile based Information System," presented at the
23rd Australasian Conference on Information Systems (ACIS 2012),
Geelong, 2012.
[39] T. Ginige and D. Richards, "Development of Mobile-based
Empowerment Processes for Sri Lankan Farmers," presented at the
24th Australasian Conference on Information Systems Melbourne,
2013.
[40] L. D. Silva, J. Goonetillake, G. Wikramanayake, A. Ginige, T.
Ginige, P. D. Giovanni, et al., "Design Science Research Based
Blended Approach for Usability Driven Requirements Gathering and
Application Development," in Second International Workshop on
Usability and Accessibility focused Requirements Engineering,
Karlskrona, Sweden, 2014, pp. 17 - 24.
[41] P. D. Giovanni, M. Romano, M. Sebillo, G. Tortora, G. Vitiello, L.
D. Silva, et al., "Building Social Life Networks through Mobile
Interfaces the Case Study of Sri Lanka Farmers " presented at the IX
Conference of the Italian Chapter of AIS (ITAIS 2012), Rome, Italy,
2012.
[42] A. G. Tansley, "The use and abuse of vegetational concepts and
terms," Ecology, vol. 16, pp. 284-307, 1935.
[43] F. S. Chapin III, P. A. Matson, and P. Vitousek, Principles of
terrestrial ecosystem ecology: Springer Science & Business Media,
2011.
[44] J. Weerahewa, S. S. Kodithuwakku, and A. Ariyawardana, "The
Fertilizer Subsidy Program in Sri Lanka," in Food Policy for
Developing Countries: Case Studies, P. Pinstrup-Andersen and F.
Cheng, Eds., ed: Cornell University, Ithaca, New York., 2010, pp. 1 -
12.
[45] Finance Minister, "Sri Lanak Budget Speech 2016," Treasury, Ed.,
ed. Colombo: Government of Sri Lanka, 2015, p. Para 372.
[46] T. S. Parikh, N. Patel, and Y. Schwartzman, "A Survey of
Information Systems Reaching Small Producers in Global
Agricultural Value Chains," presented at the International Conference
on Information and Communication Technologies for
Development(ICTD), Bangalore, India, 2007.
[47] N. Patel, S. Agarwal, N. Rajput, A. Nanavati, P. Dave, and T. S.
Parikh, "Avaaj Otalo: A field study of an interactive voice forum for
small farmers in rural India," in Human factors in computing systems
USA, 2010, pp. 733-742.
[48] B. V. Ratnam, P. Krishna Reddy, and G. S. Reddy, "eSagu 1: An IT
based personalized agricultural extension system prototype – analysis
of 51 Farmers’ case studies," International Journal of Education and
Development using Information and Communication Technology
(IJEDICT), vol. 2, pp. 79-94, 2006.
[49] A. Pande, S. Kimbahune, D. K. Singh, and A. Gupta, " mKrishi:
Facilitating Farmers in Enhancing Agricultural Production,"
presented at the A compendium of Pioneering Initiatives in e-
Agriculture in India and Around the World-14th National e-
Governence Conference, Aurangabad, India, 2011.
[50] W. Karugu, "Kenya Agricultural Commodity Exchange (KACE):
Linking Small Scale Farmers to National and Regional Markets,"
New York, United Nations Development Programme2010.
[51] Esoko Networks. (2013, October 3). Esoko. Available:
https://esoko.com/
[52] mFarm. (2013). mFarm. Available: http://mfarms.org/
611