doi:10.1016/j.eswa.2004.12.012


BioMen: an information system to herbarium

M. Delgado, W. Fajardo, E. Gibaja, R. Pérez-Pérez*

Department of Computer Science and Artificial Intelligence, Universidad de Granada,

E.T.S.I. Informática. C\ Periodista Daniel Saucedo Aranda, 18071 Granada, Spain

Abstract

This paper presents the development of BioMen (Biological Management Executed over Network), an Internet-managed system. By using

service ontologies, the user is able to perform services remotely from a web browser. The services are managed by means of a Multi-Agent

System, i.e. an Input/Output system, which interacts with the web server. In addition, artificial intelligence techniques have been incorporated

so that the necessary information may be obtained for the study of biodiversity. We have built a tool which will be of particular

use to botanists and which can by accessed from anywhere in the world thanks to Internet technology. In this paper, we shall present

the problems we encountered when building this tool and how we managed to overcome them.

q 2004 Published by Elsevier Ltd.

Keywords: Information system; Botanist; Biodiversity; Multi-agent system; Semantic web; Service ontology
1. Introduction

A herbarium is defined as a place where collections of

dried, classified plants are stored before being used as

material for the study of botany. The specimens contained in

herbariums are and always have been the essential base for

performing systematic, floral and biogeographical studies;

in addition, as a collection of perfectly identified and

ordered dried plants these represent a permanent record of

biodiversity.

It is currently calculated that more than 2.5 billion

specimens are to be found in natural history museum

collections and herbariums throughout the world (Duck-

worth, Genoways, & Rose, 1993). Biological diversity

research and study requires satisfactory access to this

biological information. As this complex information is

currently distributed among herbariums all over the world,

this makes it practically inaccessible (Berendsohn et al.,

1999).

After a time when the declining interest in global

questions entailed a certain neglect of floral studies,
0957-4174/$ - see front matter q 2004 Published by Elsevier Ltd.

doi:10.1016/j.eswa.2004.12.012

* Corresponding author. Tel.: C34 95 824 9562; fax: C34 95 824 6251.

E-mail addresses: mdelgado@ugr.es (M. Delgado), aragorn@ugr.es

(W. Fajardo), gibaja@decsai.ugr.es (E. Gibaja), ramon@decsai.ugr.es

(R. Pérez-Pérez).
and consequently of the work of herbariums, the current

preoccupation for the deterioration of the environment has

resulted in new interests being defined which has again

brought to the forefront the concern for the particular

composition of our surroundings. The study and conserva-

tion of plant biodiversity is undoubtedly one of the most

important challenges facing botanists, naturalists, environ-

mentalists, etc. in the next millennium. Let us take the

example of GBIF (Global Biodiversity Information Facility)

(GBIF, 2004), an international project which aims to make

all important data about biodiversity freely and universally

available on Internet. The approach of GBIF shall contribute

to economic growth, ecological sustainability, social effects

and scientific research by increasing the usefulness,

availability and completeness of the main scientific

source of available biodiversity information on Internet.

Most of the information necessary for GBIF comes from

information stored in the herbariums all over the world.

Therefore, by its own nature, the herbarium once again

becomes an essential piece for the development of these

objectives and those in charge of it are responsible for

providing the response called for by the research

community.

Consequently, one of the prime current needs is to

acquire updated, relevant, scientifically contrasted and

easily accessible information as the basis for conservation,
Expert Systems with Applications 28 (2005) 507–518
www.elsevier.com/locate/eswa

http://www.elsevier.com/locate/eswa


M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518508
the handling and the sustainable use of biodiversity.

However, the complexity and variability of studies carried

out in this field has forced these institutions to adopt new

techniques and protocols which are capable of responding to

the ever growing demands (Berendsohn, 2001).

One first approach at tackling this problem of manage-

ment was developed by Pando (1991) with the Herbar

program. Due to the fact that the management problem is

one shared by all herbariums, individual software packages

have been developed for each institution in order to solve

their problems, and there is no single, duly developed,

standard solution. Generally speaking, the solutions

advocated by each herbarium are partial solutions for

their problems and have been developed by the herbar-

ium’s own staff with little experience of system

development.

Below, we shall present some of the most important

software packages and their features:
1.
 HERBAR (Real Jardı́n Botánico de Madrid, 1991): free

software, entirely developed in Microsoft Access. This

reflects some of the special characteristics common to

herbarium management such as for example, the entering

of information, genera, countries, provinces, information

filtering. It creates labels.
2.
 Virtual Herbarium Express (the New York Botanical

Garden, 1996): free software which enables data input by

means of forms created in Microsoft Access XP. This

software allows labels and certain reports to be created.
3.
 BRAHUS (University of Oxford, 2002): free software

developed in Visual FoxPro using DBASE as the

databases. This is one of the most complete software

packages and allows a better control of information,

printing of labels, user policy, etc.

The following characteristics are common to all of the

above software packages:
1.
 Free software.
2.
 Use of not particularly powerful database managers

(Access, Dbase).
3.
 Data entered using templates.
4.
 Information filtering.
5.
 Label creation.
6.
 Decentralized use of software.

After analyzing a herbarium’s needs, it can be seen that

the systems developed so far have not been able to

incorporate a large number of requirements. For this reason,

BioMen was developed, and by taking advantage of modern

communication technologies, the information is available

online. The center’s information can therefore be consulted

without having to be requested from the center itself, in line

with the philosophy of GBIF (2004) whereby all infor-

mation should be public knowledge. This system adminis-

ters all herbarium tasks.
In particular, this has been implemented in the prototype

stage in the herbarium at the University of Granada. This

center was chosen for its complexity and completeness. It

meets all the characteristics to be modeled and presents all

the problems to be resolved of an extremely important

center historically among herbariums. The University of

Granada’s herbarium currently belongs to the executive of

the Iberian and Macaronesian Herbarium Association.

After this brief introduction, we shall go on to explain the

work carried out. In Section 2, we shall present an analysis

of the problem. In Section 3, we shall show the

characteristics of the developed system. In Section 4, we

shall show how the information about the treatment of

biodiversity has been obtained. Finally, Section 5 ends with

a series of conclusions about the system.
2. System analysis

A herbarium, as we have already mentioned, is a place

where collections of dried, classified plants are stored, so that

these can later be used as material for botanical study. From

this definition, we can highlight the concepts of storage and

study. The stored material is studied in order to obtain

information which will be used for the conservation,

handling and sustainable use of biodiversity (Lane et al.,

2000).

The need therefore arises for the information to be

available in a suitable, standardized form so that it may be

studied by researchers. These specimens are constantly

consulted both by researchers from the same center and

from other centers, so that relevant studies can be made, i.e.

hot-biodiversity, specific richness, etc.

The specimens stored in the center come from the

collection campaigns carried out by the center and from

exchanges with other centers. Once the specimen has been

collected, it undergoes a process of preparation in folds and

of identification. The identification process reflects the

classification of the specimen and how it will be entered

into the system databases.

The specimens are lent to other researchers so that these

may carry out their research work. These loans must be

administered by the center’s administrative staff. It is

therefore necessary to be able to control the loans made and

their period of validity. Lending material entails:
1.
 The possibility of making a discovery in the samples

which the center makes available to researchers.
2.
 Time being wasted for the center staff in charge of

administering and controlling these loans and returns.

Once the material has been studied, the researchers issue

rectifications, if necessary, about its identification. Because

of the complexity of nature, there are samples for which

there are discrepancies in the identification or simply due to

changes in the way the specimen has been identified.

Consequently, the rectifications issued must be entered by



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518 509
the center’s operators, indicating that the specimen has been

updated.

Due to the value of the information in these centers, the

problem arises of restricted access for users to certain types

of information. As a result, different levels of access are

created and each user can be allocated a level thereby

enabling or refusing access to certain types of information.

Having arrived at this point, we can see a series of

characteristics which the system must offer in order for the

objectives marked by a herbarium to be fulfilled from the

administrative point of view:
1.
 Information administration (introduction of folds, revi-

sions, genera, etc.)
Fig. 1. Client/server architecture of the system.

2.
 Information consultation (folds, etc.)
3.
 Multimedia administration
4.
 User administration
5.
 Report creation
6.
 Issuing of labels
7.
 Loan control

Another of the most important points for a herbarium,

besides administration, is to be able to satisfy the demands

of biodiversity studies. These studies use databases, and this

can at times pose quite a complex task because of the way

the data is displayed. We should therefore point out that the

system must provide a series of services so that studies may

be obtained about:
1.
 Specific richness (this is the number of species in a

certain region or location)
2.
 Taxonomic complexity (complexity when it comes to

identifying a specimen)
3.
 Study of the alpha/beta/gamma diversity (diversity

within the habitats: alpha diversity; between the habitats:

beta diversity; and for all the habitats being studied:

gamma diversity) (Rosenzweig, 1995).
4.
 Orientation in the collection campaigns.

This would entail a complex treatment of the data since

this information is not directly accessible and would require

a data acquisition and processing process in order to reach

the desired objectives (Research Directions in Biodiversity

and Ecosystem Informatics, 2001).
3. System design

Among all the possibilities which currently exist to

tackle the problem, the convenience of information systems

was thought of because of the intrinsic nature of the

problem. According to Lucas (1987), we can define

information system as a set of organized procedures,

which when performed, provide information for decision-

making and/or control of the organization. The general

theory of systems on which the information system analysis
and design is based, indicates that it is necessary to consider

the system to comprise smaller subsystems. The connection

of the smaller systems with the larger systems forms a

hierarchy which is characteristic of the theory of systems. It

also shows us that we must have an overall view of the

system, knowing that all the system components are

interrelated and interdependent, with this being one of the

most important tasks.

We can therefore say that BioMen is an information

system with a client–server architecture (Fig. 1) designed

for herbarium management. Researchers and those inter-

ested in this subject matter can gain online access to a

virtual center which models the real behavior of the units

which comprise the research center, and they are able to

obtain all the information offered totally dynamically. The

virtual users request (remote and/or hybrid) services which

will enable them to perform all the intended operations

within the system.

BioMen offers a series of services which enable the users

to have:
–
 All the centralized information
–
 Security protocols
–
 Greater computational power

The services offered might be:
1.
 Remote services: remote execution of processes and

return of the results to the user.
2.
 Hybrid services: interaction between local and remote

processes, e.g. integration of barcode readers.

The majority of the services are remote, although there

are some multimedia services which will need hybrid

services (remote image processing and inclusion).

As BioMen needs a representation of the domain

knowledge, our system uses a service ontology described

by means of the DAMLCOIL terminological system, and
the services are described using OWL-S (OWL-S is an

OWL-based web service ontology). This enables us to



Fig. 2. Hierarchy of services.

M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518510
organize the services on a graph and to provide a description

of the services including the characteristics of each service

(Fig. 2).

The ontology is used by the system to enable the user to

select the desired service, provide the necessary parameters,

and the system is therefore in charge of executing the selected

service. Due to the characteristics of the system, implemen-

tation has been carried out using agent technology (Fig. 3).

In the last 25 years, we have seen the appearance of

several paradigms to design software systems such as

procedural programming, structured programming, object

orientation and component-based software. Agents (Weiss,

1999; Wooldridge, 2002) are now being championed as the

next generation paradigm to design and build complex and

distributed software systems. An agent-based architecture

provides additional robustness, scalability, flexibility, and is

particularly appropriate for problems with a dynamic,

uncertain, and distributed nature. In particular, they seem

to be the ideal computational model for developing software

for Internet, and open networked system with no single

controlling organization (Jennings, 2000). Lastly,
Fig. 3. Way system acts given user interaction.
architectures allow the incremental development of modular

systems not only because of the modular nature of the

agents, but also because of the possibility to incorporate

legacy code by wrapping it within an agent interface.

In a multi-agent system (MAS), agents interact with one

another to achieve their individual objectives by exchanging

information, cooperating to achieve common objectives, or

negotiating to resolve conflicts. Alternative flexible patterns

of interaction have been used such as the Contract Net

Protocol (Reid & Smith, 1980), where a task is advertised by

a coordinating agent and is assigned to the agent that makes

the best bid. However, details of all possible interactions

between agents cannot be foreseen a priori and consequently:
1.
 Agents need to be able to make decisions about their

interactions at run-time, and
2.
 The organizational relationships between agents need to

be represented explicitly (e.g. peer member in a team,

manager, coordinator) by means of constructs such as

roles, norms, and social laws.

An agent is anything which can be observed sensing its

environment using sensors and acting on this environment

by means of effectors/actuators. The programming language

which has been used is Java, enabling us to include a greater

number of mobile devices and operating systems.

The way the user interacts with the user is even simpler.

The user interacts with the server using the HTTP protocol,

performing the operations desired by means of a totally

pleasing interface and without needing to have any

additional tool installed. Once the web server has gathered

the user’s request, it interacts with the multi-agent system in

order to carry out the service requested by the user (Fig. 4)

and returns the results of the service. The multi-agent

system is made up of the following agents:
–
 User
–
 Request manager or coordinator
–
 Service execution agents:

i. Remote
Fig. 4. Way of acting in a remote operation.



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518 511
ii. Hybrid
The agents which form the multi-agent system use a

blackboard architecture (Hayes-Roth, 1985; Kowalski and

Kim, 1991; Nii, 1986a,b) for communication. The black-

board is implemented by a series of tables. The agents use

the blackboard to exchange the necessary information.

Having looked at the operational logic of the system, we

shall describe the characteristics of the system and how the

requirements of a system for a herbarium have been solved.

Given the security restrictions called for by the complex-

ity and the intrinsic value of the information managed by the

center, it is necessary to create at least three different levels

of access, which will match the following user types:
Fig. 5. Consultation without identification.
–
 Internet user. Anyone wishing to consult the center’s

data. The information available for this group of users is

limited and controlled by the center’s staff. This is due to

the quantity of useful information which can be misused.

For example, knowing the location of protected species

by conducting environmental impact studies without the

respective authorization. We can therefore speak of a

data user who does not receive information processed by

the system.
–
 Researcher. System user who can access more infor-

mation than the Internet user and who in turn cannot

perform any of the characteristic operations of the center

management, such as the modification of information,

insertion of loans in the system, etc. This user has access

to the knowledge obtained by the system such as the

representation of a specimen in the center, hot-biodiver-

sity, etc. obtained automatically by the system, by means

of the application of artificial intelligence techniques.
–
 Center staff. Within this role we can in turn distinguish

two different users according to their functions within the

system:

i. Operators. All the data contained in the system must

be entered by someone, this would be the case of the

operators. These will be responsible for inserting new

folds into the system, new revisions, managing the

images associated to the folds so that they can be

consulted online, issuing labels for the preparation of

folds, etc.

ii. Managers. User in charge of supervising and mana-

ging the center. This user will be authorized to access

all the parts of the system, manage the accesses of the

different users, obtain useful information for the

center, such as for example, help in the planning of

the collection campaigns by knowing which are the
least represented specimens in the center.
Fig. 6. Identified access in the system.
3.1. From the client’s point of view

As we have already mentioned, the client communicates

with the server using the HTTP protocol in order to execute

the desired (remote and/or hybrid) services. In addition,
the HTTP protocol will show the desired information and

enter the parameters.

There are two ways to access the system:
–
 Without identification (see Fig. 5). The system is

accessed as an Internet user and therefore access to the

information is severely limited.
–
 With identification (see Fig. 6). The system controls the

different authenticated user types using login/password

and allows more or less sophisticated services to be

carried out according to the level of security allocated by

the managers.

The identified users access the system from the

authorization window (see Fig. 6), beginning a new session.

Having been identified in the system, menu systems are

created which show the services allowed according to the

level of security (see Fig. 7). Through the menu systems and

the I/O interfaces, the system will receive requests and will

provide the user with the requested information.

By means of this access, whether identified or not, the

client connects to the virtual center and the number of

operations which can be performed will depend on the

authorizations granted by the center managers. In this

virtually centralized way, we allow the system user to

overcome the problems of distance and time-wasting:
1.
 Loss of time. If we need any of the material contained in

the center we will have two options:

a. Ask the center managers for a loan. These loans will

only be granted to institutions or researchers with



Fig. 7. Different options according to the user.

M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518512
a certain degree of confidence in the treatment of

specimens.

b. Visit the center so as to personally consult the desired

specimens. As a result, the second of the centraliza-

tion problems arises-that of distance.
2.
 Distance problems. The user/researcher will not always

be in the same physical place as the center where the

information needed is.

Another of the problems which exists in both options is

that only those materials which are physically available and

which have not already been lent can be borrowed or

accessed.

In order to solve these problems, the need arises to

centralize all the center’s information so that any user/

researcher can access the information (with more or fewer

restrictions) simply, quickly, and comfortably. Conse-

quently, a client/server has been used where all the

information is virtually centralized, so that any user can

consult the information needed without having to waste time

by visiting the center or waiting for a loan.
3.2. From the server’s point of view

From the server, pages are dynamically generated for

each of the users, enabling all of the services required of the

center to be performed. By maintaining a client/server

structure, we provide solutions to the location problems

which have previously been mentioned. Therefore, the

server will act as a virtual center enabling as many services

as those allowed to each user by the center managers.

All these services are carried out and are managed by the

multi-agent system totally transparently to the user so that a

dynamic system is obtained with excellent features from the

client’s and the server’s point of view.

Any system of these characteristics must present a series

of security features which guarantee the durability of

the system. Merely thinking of the operator hours needed

to enter the 130,000 records in the database from the large
quantity of intrinsic knowledge to these records, which in

their day have been provided by taxonomical experts who

have identified and checked the specimens.

For this reason, and as a safety measure, both to

guarantee access to the information and to guarantee the

safety of this in view of severe, enormous problems, the

system is replicated identically on two servers which work

with a policy of a daily update. This policy has been chosen

rather than a real-time update as in this way we can take

action if errors have been entered by the users. Errors are not

therefore propagated in real time, and so there is a better

treatment of the information.

Let us look at the description of each of the services

managed by the server. These will be accessible according

to the user’s profile. From these services, we shall jointly

deal with all the information system.
3.2.1. Access without authentication

In this case, it will only be possible to access the system

as a data user. This user will only be able to obtain from the

system a set of lists obtained from the database. From the

access without identification, any Internet user can consult

the system and obtain a set of data entirely online. Not all

the specimen information is shown as there is information

about the specimens classified in the center which due to

misuse of the information, and given its quality, could

endanger the conservation of the species referenced.
3.2.2. Authenticated access

In order to enable more important services to be

performed within the system, an identified access has been

provided. From the server, the security certificate is issued

which the user must accept before continuing with the

access. Once this has been accepted, the information is

exchanged between the client and the server using a 128-bit

SSL secure protocol. In this way, we avoid the possibility of

data being filtered by malicious people.

Once the user has been identified in the system, the

system returns the series of services permitted.



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518 513
Below we shall describe some of the more important

tasks which can be carried out within the virtual center. The

presentation of the different services is organized from

the introduction of the data in the system until the

information processed by the information system is

obtained. The following modules are described below:
1.
 Introduction of folds
2.
 Management of revisions
3.
 Label creation.
4.
 Management of images
5.
 Advanced consultation and/or consultation of folds
6.
 Management of loans
7.
 Administration
8.
 Treatment for the biodiversity

Introduction of folds

The herbarium works with plants and consequently it is

necessary for these to be treated and conserved. The container

with one or more specimens is called a fold. The fold

univocally identifies the sample in the center and can contain

one or more specimens. The information about the fold

(number of specimens, taxonomic names, collectors, exicata,

location, UTM, height, etc.) is entered into the system by

means of the corresponding service. Once the user has selected

the desired service, the set of parameters to be provided is

returned. When these parameters are entered and returned to

the system, the agent in charge performs the remote service,

returning the result of the operation to the user.

Management of revisions

Because of the interest required by other researchers for

samples from other centers, folds are lent to other centers or

researchers. The taxonomic name of a specimen is not easy

to identify due to the complexity of nature. For this reason,

there may be discrepancies between different researchers

about the taxonomic denomination given to the specimen.

Each of the researchers’ proposals, in time, is recorded in

the system as the current taxonomic denomination (the last

in the instant t). This information will be very important
Fig. 8. Sample o
when it comes to studying the taxonomic changes in order to

be able to make the biodiversity studies.

Creation of labels

Once the fold has been entered into the system, the center

staff can obtain a label which is placed on the fold, so that

part of the information and the fold identifier can be seen.

For the identification of the fold, an alphanumeric chain

has been used. Each fold is assigned a barcode (Code 39) so

that the subsequent treatment of the fold in the system will

be a much simpler process (see Fig. 8).

These labels will mostly be used for loan management.

By adding the barcodes, certain hybrid services (barcode

reader and server) can be carried out.

Multimedia management

As we have already mentioned, one of the features of the

service which distinguishes it from others is the incorpor-

ation of a multimedia service which enables the user to

obtain much more detailed information. There is a multi-

media element associated to each fold, for example, fold

images, video of the habitat, etc. which any authorized user

can consult. This service would require the execution of

local services (reproducer, image viewer, etc.).

From the point of view of information management and

incorporation, users can pre-process the information which

they wish to incorporate for a given fold. For example,

scanning the center’s folds and associating an identifier with

the scanned image, recording a video about the collection of

the specimen in the field, etc.

This is one of the solutions offered by the system to

improve the work of the center’s staff. This multimedia

service is available to all system users. In this way, the

multimedia consultation of a specimen is made possible

without the need for a corresponding loan request. Therefore,
1.
f a
The researcher can consult the specimen’s multimedia

information the moment the center’s staff carry out the

operation in the system.
2.
 There is a reduction in the number of loans which the

center must make to the researchers.
label.



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518514
3.
 In turn, there is a better conservation of the center’s

material.

Advanced consultation and/or consultation of folds

If we make a consultation using the fold’s identifier, we will

only obtain information about that fold, but what happens with

the information contained in these? From the researcher’s

point of view it is much more necessary to be able to make a

consultation using the information contained within a fold

than for the existence of a fold. For example, existing

specimens for a UTM and/or above a certain height, etc.

This is, as we have already mentioned, the main

distinction between a library and a herbarium, in that the

information which is useful to the researcher is the

information which there is within the fold and not the fold

in itself. It is therefore as if we were asking about the

information contained in each book in a library. From this

service, any type of information existing in the system can

be searched for, and the result can be obtained both in

HTML and PDF so that it can be easily exported. The online

access to the information when consultations are made

bestows the power that the herbarium staff and researchers

need.

From the advanced consultation, the center’s staff can

begin to make a loan to the center requesting it. If the center

needs to lend all the folds containing the specimen Pinaceae

Pinus baciano, in a normal, non-computerized process, the

staff would need to go to the storeroom, look through the

folds one by one in order to select those requested by

the researcher. If the system is used, the fold identifiers

containing this specimen can be obtained and in turn, the loan

service of the system can be activated merely by entering the

loan recipient’s data and recording this in the system.

Loan management

As we have already mentioned, the center’s material is

vitally important for everyone, center and researchers alike,
Fig. 9. User adm
and it is therefore essential to control the loans made so that

these are returned within the stipulated period.

Therefore, the loan service in turn carries out the

following services:
1.
inis
Loan registration. The loan recipient and all the

materials lent are recorded in the system. This new

loan has an identifier within the system so as to be able to

control the loan in consecutive processes.
2.
 List of loans. List of loaned material for a loan identifier.
3.
 Loan return. This enables the complete or partial return

of the loan to be entered into the system. From this

section, we control which of the loaned folds have not

been returned.
4.
 Loan control. Control of the validity period of the loan.

In this way, we will be able to demand that the loan be

returned. It is possible that this operation is performed

automatically by the system once the return period has

ended and if there is no record of the return. The system

sends an e-mail to the loan recipient informing them of

the need to process the return of the loan.

Administration

Another just as important service is that of system

administration. Given the architecture of the system and the

way in which it is accessed, it is clearly necessary to have an

exhaustive control of the operations allowed to each user

within the system.

The system contains a list of operations and users. Using

the checkbox, the system’s service administrator can control

the services which these can make in the system (Fig. 9).

From this service, we can also request services to import

data from other databases, user registrations/cancellations,

etc.

Treatment for biodiversity

By means of a series of remote services, the user can

request information about:
tration.



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518 515
1.
 Taxonomic complexity (Magurran, Moreno) (Halffter

et al.)
2.
 Specific richness (Magurran, Moreno).
3.
 Orientation in collection campaigns
4.
 Study of the alpha/beta/gamma diversity

These remote services show the user the desired

information, using the information provided by other agents

who are constantly processing the data contained in the

databases.

We shall now see how this information can be obtained

and we solve the existing problems.
4. Biodiversity

As we have mentioned already, there is a large amount of

interesting information in the center’s databases. This

information enables important improvements to be made

in the quality of botanists’ work.

Nevertheless, the information is not usually directly

accessible since it needs to be processed from the database.

As a first approach to the solution of this problem, we can

recover and process information in order to obtain new

knowledge and determine:
1.
 Taxonomic complexity (Magurran, Moreno) (Halffter

et al.)
2.
 Specific richness (Magurran, Moreno).
3.
 Orientation in collection campaigns
4.
 Study of the alpha/beta/gamma diversity

The main disadvantages of obtaining the taxonomic

complexity are:
1.
 Existence of a large volume of data
2.
 Redundancies present in the information
3.
 The existence of synonyms in the database

Because of these problems, it is not possible to perform

the taxonomic complexity studies directly. In order to look

at this problem in more detail, we shall consider the

following example.

Below we shall show the identification of a small sample

of the specimens contained in the database. The specimen’s

name, in this case, comprises the family, genus and species:
–
 Cruciferae Alyssum spinosum
–
 Cruciferae Hormathophylla spinosa
–
 Cruciferae Ptilotrichum spinosum

If we want to know the number of different specimens,

when the count is made in the database, we would obtain

three specimens. However, according to Flora Ibérica

(1996), the three names refer to the same specimen

(Cruficerae Hormathophylla spinosa). In addition, the

order by which the name (identification) has evolved
(Alyssum/Ptilotrichum/Hormathophylla) is established.
For this reason, as we mentioned before, there are synonyms

in the database. This makes it impossible for us to obtain the

information necessary for biodiversity studies (different

number of specimens in one area, e.g. for specific richness

studies).

Below, and in view of the importance which the problem

of synonymy has within the research center, we shall

attempt to resolve the problem. In order to do so, there are

two possible courses of action:
1.
 By creating a synonym database. This alternative

accelerates the processing work. However, it offers a

series of drawbacks:

a. The size of the synonym table is very large, since

there is a great variety of species.

b. The table would have to be compiled by an expert.

The expert would have to carry out a repetitive and

tedious task.
2.
 By studying the evolutions. The name we give to the

change in the denomination of a specimen is evolution.

We shall explore this in greater depth later. This task can

be carried out without an expert having to intervene and

enables us to obtain the sequence of the change in the

identification.

Another piece of extremely interesting information

relates to orientation in the collection campaigns. This

provides the center with advantages both in terms of finances

and documentation. The idea is to provide information about

the types of specimens needed for the center to be complete

and well represented. For example, if the number of

specimens in the center is low, it might be that:
1.
 There really are few specimens.
2.
 The specimen has been lent to other research centers.

It is therefore necessary to inform the center of the

specimens which need to be collected so that the center is

complete and well-represented. The information might be:
1.
 Specimens which are not particularly represented and/or

below minimum levels.
2.
 The best path to follow in order to collect the specimens.

In order to obtain this knowledge, three intelligent agents

have been used (according to Wiener’s definition of intelli-

gence) which will act in turn within the multi-agent system

described above. These agents would constantly be observing

the media (databases) (Konolige, 1982), acquiring and proces-

sing the information in order to achieve the necessary

information. The agents deposit the information in the system,

using the blackboard, so that the users who so desire can access

it by means of the previously described corresponding services.

The first of the agents, called the revision agent, is

responsible for studying all the revisions for a specimen.

This result is taken advantage of by the agent called



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518516
the specific richness agent. This agent obtains the set of

synonyms contained in the database. This information is

necessary in order to count the different specimens which

there are in the database. In turn, the information obtained

by the two previous agents is used by the agent called the

collection campaign orientation agent. This agent issues a

report of those little represented specimens in the center.

4.1. Description of the agents

The main innovation of the proposed solution is the

treatment of the synonyms existing in the database. This is

an important problem and one covered by an international

project—Species 2000. In order to solve this problem

locally, we shall study the concept of synonymy and how

this can be solved. The method we propose does not need

the expert to intervene directly as it is based on the study of

the history of the specimens contained in the database.

In order to identify the synonyms, we use the concept of

evolution. We shall define evolution as the change in the

identification of a specimen. This evolution means that

specimens are stored in the database with different names

(i.e. synonyms). This means that the possibility of counting

how many different species there are in the database is a

complicated task without any real values.

In order to obtain the correct number of different

specimens, we use the following agents.

4.1.1. Revision agent

This agent is responsible for studying the number of

revisions which each specimen has. In this way, the material

most studied by botanists is obtained. Once we have the

number of revisions for each specimen, we can conduct the

relevant evolution studies. The agent places this information

on the blackboard. The AgenteRevisiones table is used to

enter the results.

4.1.2. Taxonomic complexity and specific richness agent

Information about taxonomic complexity

The solution to the synonymy problem was found by

studying the evolutions which a certain specimen under-

goes. Due to the formal complexity of the evolution and

synonymy problems, a botany-based complexity which is

therefore outside the field of this work, we shall present the

concept using an example:

The different blocks marked as A, B, C, D and E are the

different identifications which a specimen undergoes.
The arrows indicate the evolution to another identification.

Let us examine the example more closely.

Specimen 1 was identified as A but later, after being

researched by other researchers, it has been identified as B, C

and D. These identifications are treated as Evolutions and

therefore, the identifications A, B and C are synonyms of D.

These values are entered in the EvolutionAlert table. This

table is managed by the agent. The following information is

entered:
Specimen 1: A/D, evolution 0.

Specimen 1: B/D, evolution 0.

Specimen 1: C/D, evolution 0.

The agent enters the following fields:
1.
 Taxon. This indicates the number of the specimen. In this

case 1.
2.
 Antecedent. Previous names of the specimen. For

example: A, B or C.
3.
 Consequent. This indicates the name of the evolution. In

this case D.
4.
 Evolution. This field can take two values:

a. When evolution is 0. This indicates that the specimen

will not be identified in another way.

b. When evolution is 1. This indicates that if the

specimen is studied, it may change its identification

to that indicated by the Consequent field. This enables

us to inform the botanist of the specimens to be

revised so that the center’s material is completely up-

to-date. This information is based on the revisions

which have already been made to other specimens.
Below we shall place this information on the blackboard,

in particular in the taxones table, so that it may be consulted

by other agents. In this way, we shall enter the following

taxon in this table:
Specimen 1: D.

Specimen 2 is a synonym since there is another specimen

which has changed from state C to state D.

Therefore, we would enter in the AlertasEvolución table:
Specimen 2: C/D, evolution 1.

In this case, the evolution value of 1 indicates that if

specimen 2 were studied, it would certainly be identified as

D. Therefore, we enter it as a possible evolution.

The following specimen, specimen 3, has some revisions

identified as A and B. As in the AlertasEvolución table,

we have stored the fact that state B can become state D, due to

the sequence of evolutions which specimen 1 has. We

therefore add the following tuple to the alertasEvolución

table:
Specimen 3: B/D, evolution 1.



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518 517
We should remember that in the taxones table there is a

single tuple, indicating that of the specimens studied we

only have one different taxon.

In the following specimen, number 4, we can see that as

state D is revised to state E, state D becomes a synonym of

E, and it is therefore necessary to revise the data stored in

the AlertasEvolucion table and in turn, to update the

identified taxa. As a result, the tables would remain as

follows:

AlertaEvolución table taxones table

Specimen 1: A/E, evolution 0 Specimen 4: E
Specimen 1: B/E, evolution 0
Specimen 1: C/E, evolution 0
Specimen 1: D/E, evolution 1
Specimen 2: C/E, evolution 1
Specimen 3: B/E, evolution 1
Specimen 4: D/E, evolution 0
Having taken a general look at the example, we shall use

the data which we have shown previously in order to see

how the agents would act:

Fold GDAC2745:
Revision 1: Cruciferae Ptilotrichum spinosum (L.) Boiss.
Revision 2: Cruciferae Hormathophylla spinosa (L.)

Küpfer

Fold GDA28909:
Revision 1: Cruciferae Alyssum spinosum L.
Revision 2: Cruciferae Alyssum spinosum L.
Revision 3: Cruciferae Ptilotrichum spinosum Boiss.

When the multi-agent system acts, we would obtain the

following final situation.

AlertaEvolución table:
–
 Fold GDA28909: Cruciferae Alyssum spinosum/Cru-
ciferae Hormathophylla spinosa, evolution 0.
–
 Fold GDA28909: Cruciferae Ptilotrichum spinosum/
Cruciferae Hormathophylla spinosa, evolution 1.
–
 Fold GDAC2745: Cruficerae Ptilotrichum spinosum/
Cruciferae Hormathophylla spinosa, evolution 0.

Taxa table:
–
 Fold GDAC2745: Cruciferae Hormathophylla spinosa
As we can see, we obtain the desired result without

needing to produce any table which contains these

synonyms and which would involve a great deal of work

for the specialist. So far, all the solutions provided for the

synonymy problem have involved the construction of the

synonym table Species2000. We therefore believe that we

have provided an easy and innovative way for the researcher

to obtain very important information which does not entail

any expense for the center using it. This information

determines the taxonomic complexity and the richness of

species for any area and consequently enables biodiversity

studies to be made.

Information about specific richness

Once the synonym problem has been solved, we can

study the specific richness of the areas covered by the

center. In this way, botanists have a large amount of useful

information for their hot-biodiversity studies.

In turn, the information which the specific richness agent

provides can be crossed with the previously mentioned
project macro, Species2000. This project attempts to solve

the problem of synonymy as a whole. If we cross the

information developed by the agent with the information

provided by Species2000, the category which a research

center has can be established, indicating the number of

series of species which this center covers. Suitably planned

centers can therefore be combined and in accordance with

the needs of the project to be embraced and covering the

desired species.
4.1.3. Collection campaign orientation agent

If we combine the information obtained by the existing

agents, we can obtain yet more advantages. We obtain the

necessary specimens to be collected so that we may have a

complete center. The systems issues reports filtering the

synonymy problems.

In order to achieve a complete center, both on a

geographical level and in terms of plant groups, we cross

the information obtained by the agents with a Geographical

Information System. This fact offers a number of

advantages which will enable the center to obtain better

results from the information provided by the multi-agent

system.



M. Delgado et al. / Expert Systems with Applications 28 (2005) 507–518518
When the crossed information is obtained with a

geographical information system, we obtain different ways

of solving the problem of having a complete center:
1.
 Obtaining information from the areas which have not

been visited for collection, thereby enabling each

geographical point to be reflected in our center.
2.
 To obtain the necessary routes and itineraries in order to

collect specimens which are little represented in the

center.
5. Conclusions

In this paper, we have described our experiences of

constructing BioMen, an information system executed on

Internet and developed for herbariums. The constructed

system incorporates all the center’s needs, uses a multi-

agent system which makes the system much more dynamic

and easy to maintain. In addition, this is done entirely

independently of the user who does not need to know how

an ontology operates or how the agents must communicate

with one another.

BioMen is a totally operational system which uses the

newest technologies:
–
 Access to the system by means of a web browser.
–
 Javae Servlet technology (Sun Microsystems Corpor-

ation, 2001; Hall, 2001)
–
 Apache Server (The Apache Software Foundation,

2001a)
–
 Apache Jakarta Proyect Tomcat (The Apache Software

Foundation, 2001b)
–
 Java Agents (Borland Corporation, 2001; Eckel, 2000).
–
 JDBC for communication with the databases.
–
 Any JDBC-compatible database manager (MySql,

Oracle).
–
 Semantic web and ontologies. DAMLCOIL and OWL-S

The majority of software tools are free and this makes the

system much more attractive and enables it to be more

standardized.

In this document, we propose a solution to the problem of

studying hot-biodiversity. By means of artificial intelligence

techniques, such as agents, complex and extremely inter-

esting tasks have been performed such as the study of hot-

biodiversity, cost minimization and orientation in the

collection campaigns.

We are currently working on transferring the concept of

collection from the virtual center where it is not necessary

for there to be a physical place to store the specimens,

thereby creating a totally virtual center. In order to do so, the

latest technologies are being studied: PDA’s, GPRS, GPS,

multimedia transmission using mobile devices, etc.
References

Borland Corporation. (2001). JBuilder Enterprise 6.0.438.0 Documen-

tation. Available at: http://www.borland.com, last visited December

26th 2003.

Castroviejo, et al. (1996). Flora Ibérica. Real Jardı́n Botánico, CSIC, Vol.

II. Cruciferae-Monotropaceae pp. 193–195.

Eckel, B. (2000). Thinking in Java2nd ed. Prentice-Hall, Englewood Cliffs,

NJ.. Available at: http://planetpdf.com/.

Halffter, Moreno y Pineda, Manual para evaluación de la biodiversidad en

Reservas de la Biosfera, MT and SEA.

Hall, M. (2001). Core Servlets and JavaServlet pages. Englewood Cliffs,

NJ: Sun Microsystems Press/Prentice Hall.

Hayes-Roth (1985). A blackboard arquitecture from control. Artificial

Intelligence, 26(3), 251–321.

Henry, C. & Lucas, J. R. (1987). Sistemas de Información, Análisis, Diseño

y Puesta a punto. Paraninfo.

Jennings, N. R. (2000). On agent-based software engineering. Artificial

Intelligence, 177, 277–296.

Konolige. (1982). A first-order formalization of knowledge and action for a

multi-agent planning system.

Kowalski, K. (1991). A metalogic programming approach to multi-agent

belief. In V. de Lifschitz (Ed.), Artificial intelligence and mathematical

theory of computation: Papers in Honor of John McCarthy (pp. 231–

246). Boston, MA: Academic Press, 231–246.

Lane, M. A., Edwards, J. L., & Nielsen, E. (2000). Biodiversity informatics:

the challenge of rapid development, large databases, and complex data

(keynote). In VLDB 2000, Proceeding of the 26th international

conference on very large data bases.

Magurran. Diversidad. Ecologı́a y su Medición.

Moreno. Métodos para medir la biodiversidad, MT & SEA.

Nii (1986a). Blackboard systems: The blackboard model of problem solving

and the evolution of blackboard architectures.

Nii (1986b). Blackboard systems (part two): Blackboard application

systems, blackboard systems from a knowledge engineering

perspective.

Pando, F. (1991). El Herbario de Criptógamas del Real Jardı́n Botánico y

sus bases de datos. IX Simposio Nacional de Botánica Criptogámica,

Salamanca.

Reid, G., & Smith, T. (1980). The contract net protocol: high load

communication and control in a distributed problem solver. IEEE

Transactions on Computers, 29(12), 1104–1113.

Research Directions in Biodiversity and Ecosystem Informatics. (2001).

Report of an NSF, USGS, NASA workshop on biodiversity and

ecosystem informatics.

Rosenzweig, M. L. (1995). Species diversity in space and time. Cambridge:

Cambridge University Press.

Species2000, Available at: www.species2000.org, last visited March 26th

2004.

Sun Microsystems Corporation (2001). Servlet Specification v 2.3.

Available at: http://www.sun.com/servlet/, last visited December 26th

2003.

The Apache Software Foundation. (2001a). Apache HTTP Server Version

1.3.29 Documentation. Available at: http://www.apache.org, last visited

December 26th 2003.

The Apache Software Foundation. (2001b). Apache Jakarta Project Version

4.0 Documentation. Available at: http://jakarta.apache.org, last visited

December 26th 2003.

Weiss, G. (1999). MultiAgent systems. A modern approach to distributed

artificial intelligence. Cambridge, MA: MIT Press.

Wooldridge, M. (2002). An introduction to multiagent systems. New York:

Wiley.

http://www.borland.com
http://jakarta.apache.org
http://jakarta.apache.org
http://jakarta.apache.org
http://jakarta.apache.org
http://jakarta.apache.org

	BioMen: an information system to herbarium
	Introduction
	System analysis
	System design
	From the clients point of view
	From the servers point of view

	Biodiversity
	Description of the agents

	Conclusions
	References