BIOscrabble

Extraction of Biological Analogies out of Large Text Sources

Maria Katharina Kaiser, Helena Hashemi Farzaneh and Udo Lindemann

Institute of Product Development, Mechanical Engineering, Technische Universität München,

Bolzmannstrasse 15, 85748 Garching, Germany

1 STAGE OF THE RESEARCH

The aim of the research presented here is a support

for solution search in bio-inspired design. More

specifically, the engineer shall be supported in

extracting biological analogies out of large text

sources whereby the biological analogies must have

the potential to solve given technical problems. To

reach this aim, an approach has been developed by

Kaiser (Kaiser, 2012) which in the first instance

focuses on

 the large text source that shall be scanned and

 the search terms that shall be used to describe the

technical problem.

As the proposed text source – biological research

articles – is large and, therefore, the extraction of

useful analogies is challenging, the approach

proposed by Kaiser (Kaiser, 2012) needs further

development. Research started on how to manage

large amounts of information. In contrary to the

research on existing support for bio-inspired design,

research on managing large amounts of information

is still in the early stages. Initial case studies have

been conducted in order to test the applicability and

practicability of the existing approach and to

understand how to support information management

in bio-inspired design.

2 RESEARCH PROBLEM

The voyage of discovery is not in seeking new

landscapes but in having new eyes.

Marcel Proust

The question of how the huge amount of already

recorded biological information can be discovered

and used for developing innovative technical

solutions has been driving engineers for centuries. In

the recent decades, research activities in the field of

bio-inspired design intensified. To effectively

support the exploitation of existing biological

solutions for designing technical products, two main

questions can be addressed: 1) How can a promising

biological solution be identified by an engineer who

does not have a biological background? and 2) How

can this engineer transfer a promising biological

solution into a technical concept or product?

This research focuses on answering the first

question. In literature, two approaches can be

distinguished. On the one hand, databases have been

established to provide the engineer with edited and

often simplified biological knowledge

(http://www.asknature.org/, (Chakrabarti, 2005);

(Löffler, 2008); (Gramann, 2004); (Hill, 1997). On

the other hand, researchers tried to find ways to

support the engineer in effectively using biological

knowledge that is available in text format (research

on this area is discussed in detail in Sec. 5). In

contrast to databases, the latter approaches do not

face challenges such as the need of being initially

filled and kept up-to-date or the risk of accessing

subjectively misinterpreted biological data.

Therefore, the authors decided on providing further

support in this research area.

Searching biological analogies in biological

literature is not trivial: a search term-based search in

these sources can display an enormous amount of

biological information. It is therefore challenging for

an engineer to identify the right biological analogy

for a given technical problem. This is especially true

for searching in biological research articles.

Nevertheless, this search source is very

comprehensive and represents the current state of

biological research and is hence worth further

consideration.

3 OUTLINE OF OBJECTIVES

This paper illustrates an approach – called

BIOscrabble – that aims at supporting solution

search in bio-inspired design or, to be more precise,

at supporting the search term-based extraction of

Kaiser M., Hashemi Farzaneh H. and Lindemann U. (2013).

BIOscrabble - Extraction of Biological Analogies out of Large Text Sources.

In Doctoral Consortium, pages 10-20

 SCITEPRESS

relevant biological analogies out of large biological

text sources. The objectives of this paper are:

 to present the approach BIOscrabble that

- supports the engineer in performing search

term-based searches in large biological text

sources. The engineer is advised to choose

search terms that describe the technical

problem entirely and aptly.

- proposes prospects for supporting the engineer

in managing a large number of search results

by using a graph-based representation of the

search.

 to discuss the implications of the application of

BIOscrabble for the search process and the

organization of the search results.

4 STRUCTURE OF THIS PAPER

In Sec. 5 an overview on related research activities

in the field of discovering biological analogies in

text sources is given. Sec. 6 and 7 illustrate the

BIOscrabble approach and the BIOscrabble software

prototype. The application of BIOscrabble is

discussed in Sec. 8. The discussion is based on

findings from case studies in which BIOscrabble

was applied to technical problems. The expected

outcome of the BIOscrabble research project is

illustrated in Sec. 9. The paper concludes with an

outlook on future work in Sec. 10.

5 STATE OF THE ART

Research activities have been conducted that focus

on the question of how to effectively discover

biological information available in text format. In

the following, an overview over research activities is

given that address the issues of choosing text

sources and search terms, overcoming linguistic

differences between engineers and biologists and

identifying relevant biological analogies.

5.1 Text Sources

Which text sources have been proposed for an

effective discovery of biological information

available in text format so far?

A natural language approach to biomimetic

design has been developed at the University of

Toronto. Keyword searches were performed and

analysed in an introductory biological textbook to

support an effective retrieval of biological

information available in text format. The textbook

was chosen as an initial text source, because 1) it is

intelligible for a reader without a biological

background and 2) it provides biological information

ranging from molecular structures to ecosystems

(Shu, 2010); (Cheong, 2012).

Biological textbooks were also used by Stroble et

al. and Nagel et al. (Stroble, 2009); (Nagel, 2010)

for compiling engineering-to-biology thesauri and

by Nagel and Stone (Nagel, 2011) for testing a

methodology for facilitating systematic biologically

inspired design (including searching for biological

analogies).

These textbook-based approaches can be applied

to any other text source. However, advanced sources

such as biological research articles are suggested for

finding further details on selected biological

analogies rather than for initial searching (Shu,

2010).

The World Wide Web was analysed as an initial

source from which biological inspiration can be

drawn by Vattam and Goel (Vattam, 2011).

Searching the web for relevant biological analogies

was found to be very costly in terms of time spent on

searching and analysing (resource costs). The

probability of spending time on analysing useless

information or of ignoring useful information due to

uncertainty was found to cause additional costs

(opportunity costs). Using the web as an initial

source for searching for biological analogies was

also explored and supported by Vandevenne

(Vandevenne, 2011; 2012b). The approaches are

further described in Sec. 5.4.

The approaches described in Sec. 5.1 did not

focus on using biological research articles as an

initial search source for discovering biological

information available in text format. As biological

research articles are representing biological research

most comprehensively, they are used as an initial

search source in the approach proposed here.

5.2 Search Terms

What types of search terms have been proposed for

an effective discovery of biological information

available in text format so far?

To search biological texts for relevant biological

analogies, at the University of Toronto, primarily

verbs describing the desired technical function or

effect were used as search terms (Shu, 2010);

(Cheong, 2012). In one work the use of adjectives

describing the desired technical qualities was

illustrated (Ke, 2010). Nouns were found to indicate

preconceived solutions and were therefore neglected

BIOscrabble-ExtractionofBiologicalAnalogiesoutofLargeTextSources

(Shu, 2004; 2010).

Functional technical terms or terms of the

Functional Basis (widely accepted characterization

of product functions in a verb-object (function-flow)

format (Stone, 2000) serve as basis for the biological

terms in the engineering-to-biology thesauri

developed by Stroble et al. and Nagel et al., (Stroble,

2009); (Nagel, 2010). The engineering functional

terms contained in the thesauri play a central role in

searching for relevant analogies in the methodology

proposed by Nagel and Stone (Nagel, 2011) which

aims at facilitating systematic biologically inspired

design.

For identifying biological analogies, functions or

functional verbs of technical or biological systems

are central in the work of Vandevenne (Vandevenne

2011).

It can be concluded that the approaches

described in Sec. 5.2 mainly focus on system

functions to discover biological information

available in text format. In the approach proposed

here, search terms describing system properties and

environmental influences on systems are

additionally focused on.

5.3 Terminology

What options have been revealed to bridge the gap

between an engineer’s and a biologist’s terminology

so far?

At the University of Toronto, two alternatives

were revealed which can mitigate linguistic

differences between engineers and biologists. First,

it was found that the chance of identifying relevant

biological analogies in biological text sources

increases when the engineering functional search

terms are expanded by term variations such as

synonyms (Vakili, 2001); (Hacco, 2002). Second, a

method for identifying biologically meaningful

keywords was developed (Chiu, 2005; 2007).

Textbook words collocated with the functional

search terms initially used for searching were

defined as biologically meaningful keywords

dependent on their occurrences in further text

sources. Biologically meaningful keywords were

furthermore defined for the terms of the Functional

Basis (Cheong, 2008).

To link engineers’ and biologists’ terminology

and, therefore, sharpen search results when

performing a key word search, engineering-to-

biology thesauri have been developed by other

research groups (Stroble, 2009); (Nagel, 2010).

Biologically meaningful keywords or keywords

contained in the engineering-to-biology thesauri are

only available for a portion of functional search

terms. Therefore, in this work common term

variations such as synonyms are used for varying

search terms.

5.4 Identifying Biological Analogies

What decision guidance has been given for

effectively identifying relevant biological analogies

described in text format so far?

Researchers at the University of Toronto

approached this question as follows: 1) Hacco and

Shu detected characteristics of biological texts that

contain irrelevant biological analogies (Hacco,

2002). When searching for biological analogies,

these characteristics can form the basis for a

systematic removal of irrelevant search results and,

therefore, assist the engineer in identifying relevant

analogies. 2) A characteristic of biological texts that

can effectively be used as design stimuli is the

explicit description of biological principles (Mak,

2004). A case study showed that the effort for

selecting relevant biological systems is reduced

when the principle has not to be abstracted from

described behaviors or forms. 3) To further support

the engineer in identifying relevant analogies,

different methods for categorizing biological

information have been developed based on semantic

relations (Ke, 2009); (Cheong, 2012); (Son, 2012).

Categorized biological information was found to

assist the engineer in selecting analogies relevant to

the design problem (Ke, 2009); (Cheong, 2012).

Further support for identifying relevant

biological analogies was provided by Vandevenne

(Vandevenne, 2011; 2012a; 2012b) in different

ways: 1) To support the identification of relevant

analogies, textual descriptions of biological systems

that are detectable in the web were automatically

characterized and mapped to technical problem

descriptions (Vandevenne, 2011). 2) To support the

structuring and, therefore, the selection of biological

information available in text format, Vandevenne et

al., introduced a method for classifying biological

analogies into the Biomimicry Taxonomy of the

Biomimicry Design Portal Ask Nature

(http://www.asknature.org/, Vandevenne, 2012a). 3)

For filtering out and preselecting biological

information that is available in the web, a

webcrawler was developed that continuously

collects documents containing biological

information that is relevant to biomimetics

(Vandevenne, 2012b).

In the approaches described in Sec. 5.4, no

assistance for identifying biological analogies

IC3K2013-DoctoralConsortium

described in biological research articles was

provided explicitly. Nevertheless, concepts such as

the categorisation or clustering of biological

information are used in the approach proposed here.

6 BIOscrabble – APPROACH

As with the approaches described, BIOscrabble

supports the effective discovery of biological

information available in text format. The main

differences to the existing approaches described in

Sec. 5 are: 1) the exclusive use of biological

research articles as an initial search source, 2) the

explicit inclusion of search terms that describe a

desired technical system’s properties and the

environmental influences on this system and 3) the

use of a graph-based representation of the search

that has been carried out for assisting the engineer in

identifying relevant biological analogies.

In the following, the BIOscrabble approach and a

BIOscrabble software prototype are illustrated.

BIOscrabble is a further development of the

approach proposed by Kaiser (Kaiser, 2012). In the

following a short overview over BIOscrabble is

given. Further descriptions of the BIOscrabble steps

and components as well as the reasons for their

inclusions are given in Sec. 6.2 – 6.5.

6.1 Overview

For the effective discovery of biological information

available in text format, the following steps are

proposed in BIOscrabble (Fig. 1):

1) The user – here the engineer – describes the

technical problem under consideration in terms

of system functions, system properties and the

environmental influences on it (definitions see

below). The search terms for the following

search process are taken from this description.

2) The search terms are varied, e.g. synonyms are

built. WordNet (description see below) provides

assistance here.

3) The database PubMed (description see below) is

searched using the original and the varied search

terms.

4) The PubMed search results, i.e. the biological

research articles are clustered and displayed to

the user in the form of structured graphs.

6.2 Text Sources

BIOscrabble supports using biological research

articles as initial text sources when searching for

biological analogies via search terms.

Figure 1: BIOscrabble search process.

BIOscrabble-ExtractionofBiologicalAnalogiesoutofLargeTextSources

In BIOscrabble, PubMed is used as a search

source (http://www.ncbi.nlm.nih.gov/pubmed/).

PubMed is a meta-database (i.e. a database of

databases) that refers to more than 22 million

citations for biomedical literature. Besides citations

in the field of biomedicine, health and

bioengineering, life sciences, behavioural sciences

and chemical sciences are addressed. PubMed was

developed and is maintained by the National Center

for Biotechnology Information (NCBI). It can be

freely used. PubMed can be searched by search

terms. Search terms can be connected by Boolean

operators such as AND or OR. Connecting search

terms with AND results in citations that must

contain all search terms connected. Connecting

search terms with OR results in citations that must

contain at least one of the search terms connected.

PubMed provides different filters that can be set

when performing a search. The BIOscrabble user is

advised to set the filter “abstract available”. This

filter secures that at least an article’s abstract is

available in addition to its title. This is the minimum

requirement for the identification of relevant search

results.

Biological research articles were chosen as an

initial search source, because this text source is very

comprehensive and represents the current state of

biological research. In biological textbooks, not all

biological research findings are illustrated or

illustrated in detail. This was stated in interviews

with biologists that were conducted within a project

aiming at the development of a communication

platform for engineers and biologists.

6.3 Search Terms

BIOscrabble advises to search for biological

analogies in biological research articles by search

terms that describe a desired technical system’s

functions, properties or the environmental influences

on this system. This search schema encourages the

engineer to see the considered technical problem or

system from different angles which can increase the

diversity of the search terms. This, in turn, is

supposed to enhance the probability of detecting

relevant biological analogies.

6.3.1 Functions

Search terms describing technical functions are

included in BIOscrabble because of two reasons: 1)

Existing research dealing with the discovery of

biological analogies in biological text sources

showed that good results can be obtained by using

functional search terms (Shu, 2010, Vandevenne,

2011, Cheong, 2012). 2) Functions play a key role in

modelling technical systems (Stone, 2000, Pahl,

2007, Erden, 2008). Engineers are therefore trained

in describing a technical system by its functions.

In this work, functions are defined according to

Pahl et al., (2007).

6.3.2 Properties

BIOscrabble includes search terms describing

technical system properties for the following reason:

Compared with engineers, biologists do not focus on

describing biological systems in terms of functions

as the concept of function is not an issue during their

academic training. Therefore, biological systems

which are not primarily described functionally can

be missed when searching with functional search

terms only. In a pre-study carried out by the authors,

spider silk, for example, was found to be described

in terms of system properties rather than system

functions.

Here, properties are defined according to Eder

qnd Hosnedl (2008).

6.3.3 Environmental Influences

Natural systems are strongly aligned to the

environmental influences they are exposed to as they

evolve driven by their environment. Therefore,

search terms that describe the environmental

influences on a technical system are assumed to lead

to relevant biological analogies. The above named

pre-study confirmed this assumption. Consequently,

environmental influences are included in

BIOscrabble as a search term type.

In this work, the environment is defined

according to Srinivasan and Chakrabarti (2009).

Environmental influences are hence the influences

of the environment on a technical system.

6.4 Terminology

To account for differences between an engineer’s

and a biologist’s terminology, the BIOscrabble user

is advised to vary the search terms (original search

term) he or she picks. The following search term

variations (varied search term), some of which are

supported by WordNet (see below), are included in

BIOscrabble: 1) synonyms of the search term, 2)

nouns, verbs and adjectives derived from the search

term and 3) antonyms of the search term. If the

search term would be “purifying”, 1) could be

“cleansing”, 2) would be “purification, purify and

pure” and 3) would could be “adulterating”.

IC3K2013-DoctoralConsortium

Synonyms are included as they were found to

increase the chances of getting relevant matches in

case of searching a biological textbook for biological

analogies via engineering functional search terms

(Vakili, 2001); (Hacco, 2002). The authors assume

that this is also true for searching biological research

articles with search terms describing technical

functions, properties or environmental influences.

Deriving nouns, verbs and adjectives from a search

term has the potential to broaden the solution space

without leading to results that vary from the topic.

BIOscrabble uses WordNet for automating some of

the above mentioned search term variations

(http://wordnet.princeton.edu/). WordNet is a lexical

database of English nouns, verbs, adjectives and

adverbs which are related by cognitive synonymy.

WordNet was also used at the University of Toronto

(Hacco, 2002); (Chiu, 2007); (Ke, 2010); (Cheong,

2012) for varying the terms used for searching

biological text sources for analogies.

6.5 Identifying Biological Analogies

BIOscrabble proposes to use graph-based

representations of search processes in bio-inspired

design to support the engineer in identifying

biological analogies that are hidden in large text

sources.

Graphs are abstract structures which represent

objects and their relations. Objects are represented

as nodes, relations are represented as edges. Here,

typed attributed graphs are used that contain three

kinds of node types and two kinds of edge types: the

node types are 1) original search term, 2) varied

search term and 3) research article, the edge types

are 1) edges linking search terms to their variations

and 2) edges linking search terms (original or

varied) to the research articles in which they are

contained.

Graphs showing the relations between the

original search terms and the varied search terms,

between the search terms (original and varied) and

the corresponding research articles or between

different research articles can support the

identification of biological analogies in different

ways. The engineer can work with graphs that

 cluster research articles in which the same search

terms are contained

 cluster research articles in which the same types

of search terms (function, property,

environmental influence; original term, varied

term) are contained

 cluster research articles in which the same or

different search terms are contained with a

certain frequency

 cluster research articles which are cited by each

other

 cluster research articles which are written by the

same first authors

These clusters can be supportive. Research article

samples can be taken within each cluster and

examined for relevant biological analogies. If the

samples within one cluster are promising the cluster

can be examined further. The large amount of

research articles a PubMed search generally

produces therefore is divided into manageable

“packages”. The benefit of categorizing or clustering

biological information is shown in (Ke, 2009,

Vandevenne, 2011, 2012a, Cheong, 2012).

7 BIOscrabble – SOFTWARE

PROTOTYPE

From a certain amount of search results (> 50)

onwards it is impracticable to generate these graph

views manually. Therefore, the BIOscrabble search

process including the generation of the search graph

has been automated in a software prototype.

7.1 Front End

By default the user is asked to enter search terms

describing system functions, system properties and

environmental influences on the desired system.

These different types of search terms can be

connected via check marks in order to search for

research articles that contain more than one type of

search terms. Check marks correspond to the

Boolean operator AND. Empty check boxes

correspond to the Boolean operator OR. In this case

all research articles are searched in which at least

one type of search terms is contained. To exclude

research articles containing particular terms,

unwanted terms can be added.

The user has the possibility to exclusively search

for review articles by setting the filter “Review

Articles”. Research articles which are freely

available in the full text version can be searched for

by setting the filter “Fulltext Available Articles”.

As a result of the search process, a table is

displayed. This table contains the titles and the first

authors of the resulting research articles as well as

their rank and their rating. The rank increases with

the value of the rating. Both are dependent on the

BIOscrabble-ExtractionofBiologicalAnalogiesoutofLargeTextSources

search terms’ frequency in each search result (see

7.2).

The search results can be sorted by their rank,

their title, their terms, their first authors and their

rating.

For every search process the corresponding

search graph can be viewed and saved as well as

exported as a .graphml file. In the graph, information

about the node and edge types as well as information

about the rank and rating of the resulting research

articles is contained.

Fig. 2 shows the graphical user interface (GUI)

of BIOscrabble.

7.2 Back End

After the search process has been started by the user,

the search terms which have been entered are varied.

For the variation of the search terms WordNet

Search – 3.1 is used online. Each search term is

connected to its variations by the Boolean operator

OR.

Depending on the user input, the different types

of search terms are connected by either the Boolean

operator AND or OR. The resulting search phrase is

used to search PubMed for biological research

articles. The PubMed filter “abstract available” is set

automatically for the reasons mentioned in Sec. 6.2.

The search terms (original and varied) that are

contained in an article’s title or abstract are counted.

For the generation of the search graph the

following information is extracted from the search

process: 1) which term is the origin or the variation

of which other term, 2) which term is contained in

which article and 3) how frequently is a term

contained in one article (necessary for representing

the article’s rank and rating).

With the current software prototype it is possible

to distinguish articles according to the frequency of

the search terms that are contained. As explained in

Sec. 7.1 the frequency of the search terms is

reflected in the rating and rank of the articles. The

rating is calculated as shown in equation 1:

rating = 2 * a + b (1)

In equation 1, a is the number of different search

terms contained in a research article and b is the

total number of search terms found in the article.

Figure 2: Screenshot of BIOscrabble GUI.

IC3K2013-DoctoralConsortium

Equation 1 reflects one of the observations made

during the application of the BIOscrabble approach

to different technical problems in case studies with

students of mechanical engineering (see 8.3).

Fig. 3 shows two graphs of an exemplary search.

The left graph shows the raw representation of the

relations between the search terms and the research

articles they are contained in. The right graph shows

the same relations, but the rank of the articles is

reflected in the size of the nodes representing them,

i.e. articles with a higher frequency of search terms

are represented in bigger nodes.

The implementation of other graph-based

representations (see above) is in the concept stage.

8 DISCUSSION

In the following, the approach of BIOscrabble is

discussed based on the documentation of case

studies in which BIOscrabble was applied to

different technical problems. In these case studies,

students of mechanical engineering were asked to

solve a given technical problem applying to the

search process proposed by BIOscrabble and to

document their results. They did not use the software

prototype and, therefore, were not supported in the

identification of biological analogies. For brevity,

the technical problems are not discussed in detail

here.

The proposed text source, the proposed types of

search terms including the proposed variations and

the proposed support for the identification of

biological analogies are discussed separately.

8.1 Text Sources

Biological research articles which are available via

PubMed proved to be a text source which can

provide useful biological information for the

developments of bio-inspired products. Besides

articles about biological systems, articles that

describe bio-inspired or other technical products

which are used in a biological context were found to

be useful for the solving of the considered technical

problems.

Figure 3: Exemplary graphs; left: raw representation, right: edited representation; black: node type original search term,

grey: node type varied search term, white: node type research article.

BIOscrabble-ExtractionofBiologicalAnalogiesoutofLargeTextSources

Therefore, PubMed can be viewed as a search source

that is not only useful for the development of bio-

inspired products, but can also serve as a tool for

finding analogies in general.

For most PubMed searches an impracticable

amount of research articles is displayed. Without

additional support it is very time-consuming to

identify articles that contain useful biological

information. To make a high amount of search

results manageable within a reasonable time, graphs,

as described above, can be supportive.

In conclusion, PubMed is a promising text source

for finding biological or other analogies for solving

technical problems. However, the engineer has to be

supported in exploring the huge amount of search

results. Furthermore, biological research articles are

more appropriate for getting inspiration of what

biological systems exist than for understanding the

basic principles behind them. For engineers, many of

them are hard to understand.

8.2 Search Terms

The case studies show that depending on the

technical problem different types of search terms led

to useful biological analogies.

In most cases, the best results were obtained by

searching with AND combinations of search terms

of the types function, property and environmental

influence. Nevertheless, there were technical

problems that were solved using biological systems

contained in articles that were found by only one

type of search term. Examples are the design of an

adaptive surface with variable heat conductivity, a

self-sharpening knife or a tension-reducing

mechanism for stuffed pieces of luggage. In case of

the adaptive surface, functional search terms were

most beneficial. For the knife and the tension-

reducing mechanism, search terms of the type

system property and environmental influence led to

the best analogies.

Search term variations in all cases led to

additional relevant biological analogies.

It is concluded that the types of search terms that

are proposed as well as the search term variations

are beneficial for the discovery of biological

analogies in large text sources. Whether the benefit

of certain search term types or term variations

depends on the kind of technical problem remains to

be proven.

8.3 Identifying Biological Analogies

The graph-based representations proposed in Sec.

6.5 were not available for the participants of the case

studies as they did not use the BIOscrabble software

prototype. Nevertheless, conclusions concerning a

support for identifying biological analogies can be

drawn from the relations between the search terms

that were used for searching PubMed and those

research articles that contain relevant biological

analogies.

For the development of a device for purifying

drinking water, the relevance of the research articles

correlated with the frequency of search terms that

are contained. Additionally, a slight correlation

between the relevance of an article and the

frequency of different types of search terms was

observed. These findings are realized in the paper

rating and ranking mechanism that is implemented

in the current BIOscrabble software prototype (see

Sec. 7).

For the other design case studies, this

observation did not prove true. In case of the self-

sharpening knife the most relevant research articles

contained only one single search term. Although, all

articles that were relevant for the development of a

self-sharpening knife dealt with only two topics,

namely “teeth” and “claws”. Therefore, a graph that

shows clusters of research articles that are cited by

each other or are written by the same first author –

and thus are supposed to deal with a similar topic –

is proposed.

Whether the other graphs that are proposed in

Sec. 6.5 can be beneficial remains to be proven.

9 EXPECTED OUTCOME

The expected outcome of this research project is to

provide a functional software tool for supporting

solution search in bio-inspired design. The expected

partial outcomes are the following:

 BIOscrabble will provide guidelines

- supporting the formulation of promising search

terms for finding biological analogies for a

technical problem. Depending on the technical

problem, guidelines are given which search

term type is most beneficial for searching.

- supporting the selection of term variations

according to their potential of increasing the

amount of relevant analogies.

- supporting a type of technical problem-specific

use of the graph-based representations.

 BIOscrabble will provide the possibility to

execute graph-based analyses in order support

the user in exploring the huge amount of

biological information.

IC3K2013-DoctoralConsortium

 The above mentioned partial outcomes will be

based on case studies in which BIOscrabble is

applied to a sufficiently large number of different

classes of technical problems.

 BIOscrabble will be connected to a platform

supporting the transfer of biological solutions

into technical solutions that is developed in

another research project.

10 FUTURE WORK

To achieve those outcomes, further research has to

be done and further case studies with the application

of BIOscrabble have to be performed to

 understand possible correlations between types

of search terms and relevant biological research

articles dependent on specific classes of technical

problems.

 understand possible correlations between the use

of search term variations and relevant biological

research articles dependent on specific classes of

technical problems.

 understand possible correlations between the

benefit of certain graphs and specific classes of

technical problems.

 extend the graphs that are currently proposed.

 to further prove the text source of biological

research articles (and the database PubMed), the

types of search terms, the term variations (and

the software WordNet) and the graphs regarding

their value for solution search in bio-inspired

design.

ACKNOWLEDGEMENTS

The authors want to thank the students who applied

the BIOscrabble approach in their biomimetic

development projects as well as the students who

added to the development of the BIOscrabble

software prototype.

REFERENCES

Chakrabarti A., Sarkar P., Leelavathamma B., and

Nataraju B. S., 2005, A Functional Representation for

Aiding Biomimetic and Artificial Inspiration of New

Ideas, Artificial Intelligence for Engineering Design,

Analysis and Manufacturing, 19(02), pp. 113–132.

Cheong H., Shu L. H., Stone R. B., and McAdams D. A.,

2008, Translating Terms of the Functional Basis into

Biologically Meaningful Keywords, Proceedings of

IDETC/CIE 2008 ASME 2008 International Design

Engineering Technical Conference & Computers and

Information in Engineering Conference, New York

City, NY, USA.

Cheong H., and Shu L. H., 2012, Automatic Extraction of

Causally Related Functions from Natural-Language

Text For Biomimetic Design, Proceedings of the

ASME 2012 International Design Engineering

Technical Conferences & Computers and Information

in Engineering Conference IDETC/CIE 2012,

Chicago, IL, USA.

Chiu I., and Shu L. H., 2005, Bridging Cross-Domain

Terminology for Biomimetic Design, Proceedings of

IDETC’05 ASME 2005 International Design

Engineering Technical Conferences and Computers

and Information in Engineering Conference, Long

Beach, California, USA.

Chiu I., and Shu L. H., 2007, Biomimetic Design through

Natural Language Analysis to Facilitate Cross-Domain

Information Retrieval, Artificial Intelligence for

Engineering Design, Analysis and Manufacturing,

21(01), pp. 45–59.

Eder W. E., and Hosnedl S., 2008, Design-Engineering –

A Manual for Enhanced Creativity, Taylor & Francis

Group LLC.

Erden M. S., Komoto H., Van Beek T. J., D’Amelio V.,

Echavarria E., and Tomiyama T., 2008, A review of

function modeling: Approaches and applications,

Artificial Intelligence for Engineering Design,

Analysis and Manufacturing, 22(02), pp. 147–169.

Gramann J., 2004, Problemmodelle und Bionik als

Methode.

Hacco E., and Shu L. H., 2002, Biomimetic Concept

Generation Applied to Design for Remanufacture,

Proceedings of DETC’02 ASME 2002 Design Engi-

neering Technical Conferences and Computers and

In-formation in Engineering Conference, Montreal,

Que-bec, Canada.

Hill B., 1997, Innovationsquelle Natur, Technische

Universität München.

Kaiser M. K., Hashemi Farzaneh H., and Lindemann U.,

2012, An Approach to Support Searching for

Biomimetic Solutions Based on System

Characteristics and its Environmental Interactions,

International Design Conference - Design 2012,

Dubrovnik, Croatia.

Ke J., Wallace J. S., and Shu L. H., 2009, Supporting

Biomimetic Design through Categorization of Natural-

Language Keyword-Search Results, Proceedings of

the ASME 2009 International Design Engineering

Technical Conferences & Computers and Information

in Engineering Conference IDETC/CIE 2009, San

Diego, California, USA, pp. 1–10.

Ke J., Wallace J. S., Chiu I., and Shu L. H., 2010,

Supporting Biomimetic Design by Embedding

Metadata in Natural-Language Corpora, Proceedings

of the ASME 2010 International Design Engineering

Technical Conferences & Computers and Information

in Engineering Conference IDETC/CIE 2010, Montre-

BIOscrabble-ExtractionofBiologicalAnalogiesoutofLargeTextSources

al, Quebec, Canada, pp. 1–8.

Löffler S., 2008, Anwenden bionischer Konstruktions-

prinzipe in der Produktentwicklung, Technische Uni-

versität Carolo-Wilhelmina zu Braunschweig.

Mak T. W., and Shu L. H., 2004, Abstraction of Bio-

logical Analogies for Design, CIRP Annals - Manufac-

turing Technology, 53(1), pp. 117–120.

Nagel J. K. S., Stone R. B., and McAdams D. A., 2010,

An Engineering-to-Biology Thesaurus For

Engineering Design, Proceedings of the ASME 2010

International Design Engineering Technical

Conferences & Computers and Information in

Engineering Conference IDETC/CIE 2010, Montreal,

Quebec, Canada.

Nagel J. K. S., and Stone R. B., 2011, A Systematic

Approach to Biologically-Inspired Engineering

Design, Proceedings of the ASME 2011 International

Design Engineering Technical Conferences &

Computers and Information in Engineering

Conference IDETC/CIE 2011, Washington, DC, USA.

Pahl G., Beitz W., Feldhusen J., and Grote K. H., 2007,

Engineering Design: a Systematic Approach,

Springer-Verlag, London.

Shu L. H., 2004, Generalizing the Biomimetic Design

Process, Proceedings of CDEN/RCCI Inaugural De-

sign Conference, Montreal, Quebec, Canada.

Shu L. H., 2010, A natural-language approach to bio-

mimetic design, Artificial Intelligence for Engineering

Design, Analysis and Manufacturing, 24(04), pp. 507–

519.

Son J. (Jay), Raulf C., Cheong H., and Shu L. H., 2012,

Exploring the Collective Categorization of Biological

Information for Biomimetic Design, Proceedings of

the ASME 2012 International Design Engineering

Technical Conferences & Computers and Information

in Engineering Conference IDETC/CIE 2012,

Chicago, IL, USA, pp. 1–9.

Srinivasan V., Chakrabarti A., SAPPhIRE – An Approach

to Analysis and Synthesis, International Conference

on Engineering Design, ICED09, Stanford, CA, USA

Stone R. B., and Wood K. L., 2000, Development of a

Functional Basis for Design, Journal of Mechanical

Design, 122(December).

Stroble J. K., Stone R. B., McAdams D. A., and Watkins

S. E., 2009, An Engineering-to-Biology Thesaurus to

Promote Better Collaboration , Creativity and

Discovery, Proceedings of the 19th CIRP Design Con-

ference – Competitive Design, pp. 30–31.

Vakili V., and Shu L. H., 2001, Towards Biomimetic

Concept Generation, Proceedings of DETC’01 ASME

2001 Design Engineering Technical Conferences

Design Theory and Methodology, Pittsburgh,

Pennsylvania.

Vandevenne D., Verhaegen P.-A., Dewulf S., and Duflou

J. R., 2011, A Scalable Approach for the Integration of

Large Knowledge Repositories in the Biolog-ically-

Inspired Design Process, International Conference on

Engineering Design, ICED11, Copenhagen, Denmark.

Vandevenne D., Verhaegen P.-A., Dewulf S., and Duflou

J. R., 2012, Automated Classification into the

Biomimicry Taxonomy, International Design Confer-

ence - Design 2012, Dubrovnik, Croatia, pp. 1161–

1166.

Vandevenne D., Caicedo J., Verhaegen P.-A., Dewulf S.,

and Duflou J. R., 2012, Webcrawling for a Biological

Strategy Corpus to Support Biologically-Inspired

Design, CIRP Design 2012, A. Chakrabarti, ed.,

Springer-Verlag, London, pp. 83–92.

Vattam S. S., and Goel A. K., 2011, Foraging for In-

spiration: Understanding and Supporting the Online

In-formation Seeking Practices of Biologically

Inspired Designers, Proceedings of the ASME 2011

International Design Engineering Technical

Conferences & Computers and Information in

Engineering Conference IDETC/CIE 2011,

Washington, DC, USA, pp. 177–186.

IC3K2013-DoctoralConsortium