Sewcom

From Wikipedia, the free encyclopedia

The Sewcom Method - Search the Web With Concept Maps

SEWCOM is a metacognitive method (Search the Web with Concept Maps), that makes use of concept maps to

help focalize the search, thus reducing the number of hits of the search engines. It was created and developed at

the University of Padua (Italy) - Educational Technology Center to help students to acquire competences

about Information Literacy.

This can be done by “visually” brainstorming on the lexicon to use on the search engines, and more important, to

discover a new one in the documents found, in a recursive refinement process. The last steps of the method

consist of topological re-structuration of the map in “semantic areas” that represent contextual categories.

This process can generate an interesting serendipity side-effect: it is easy to trace cross-domain links among

different semantic areas, which often results in the discovery of some important documents which would never

be found following our usual line of reasoning, and whose existence was not suspected.

Sewcom method explained

The most frequent activity carried out by teachers and students, when using Internet at school, is the search of

information by search engine (Bilal 2000, Walker & Moen, 2001). However, albeit able to use search engine

interfaces, students do not easily succeed in retrieving appropriate target information (Kafai & Bates, 1997,

Schacter et al., 1998). The reason is that they often lack metacognitive formulation of their information needs

(Eisenberg & Berkowitz, 2000; Flavell, 1977).

Concept maps (Novak, 1998) are metacognitive instruments useful to organize concepts into easily recognisable

visual structures, and here we will try to show that they can improve the Web search process in an information

literacy context. The proposed method is known as SEWCOM (Search the Web with Concept Maps ). It tries to

integrate two processes: the learning about the lexicon in the semantic domain and the learning about the best

strategies to locate the information on the Web. The method was tested in informal experiments with teachers

and students of higher schools. SEWCOM consists of four steps:

1)Brainstorming to create a conceptual map using words related to the topic to be searched.

2)A topological re-structuration (catego-rization) of the map in semantic areas and use of the search engines with

(key)words appropriate to each area to find related documents.

3)An evaluation of documents, and discovery of new lexical terms that are used to narrow down the number of

hits in a new recursive search.

4)A creative re-structuration of the mapped knowledge drawing cross-domain links.

The first two steps provide a number of con-cepts/words that will be grouped together. A "box figure" (square,

ring, rectangle, etc.) and a specific color are used to highlight words. This approach recall the Gestalt Theory.

The grouping of visual stimuli (Moore and Fitz, 1993), sorted by both proximity (objects close together are

perceived as a group) and similarity (similarly shaped objects are perceived as a group), is an important factor.

A great number of researches on cognitive psychology, particularly on semantic priming (Plaut, 2000) and on the

context effect (Lavigne, 2000) suggest that perception is favored when words are stored together, and when they

are part of a semantically homogeneous context (i.e. fork–knife-spoon). A statistical method, the [[Latent

Semantic Analysis]] (LSA), has been recently proposed in order to build semantic spaces that do not stem from

the interactive opinion of a subject but from an automatic text analysis (Landauer and Dumais, 1997). The

assumption of LSA is that similar words are encountered together in similar contexts: for example, in a part of

text concerning "cats" also "mouses" and "dogs" are likely to be mentioned. The result of the analysis can be

represented by a multidimensional space in which words present in similar context are located near to one

another.

In step 3 new searches are carried out, assuming some of the words (lexicon) previously found in documents as

key-words enabling to narrow down and better focalize the topic. During this phase some of the documents are

examined and the more relevant words found in them, if any, are added to the map. It is not easy to identify

“hot” words in documents: an appropriate selection of words, concerning both concepts and linking, is key to

represent the user’s understanding of the domain (Cañas, 2003). The choice of the words can be different not

only for different persons, as it is obvious, but also for the same person and the same document if it is done in

different moments (Furnas, 1987).

There is a double effect: on one hand, new knowl-edge is gathered by interpreting the documents that are

considered relevant; on the other, there is a consequent increasing of the specific lexicon concerning a given

topic. In fact, the learning of a new language can be improved (Stahl, 1999) especially if there is an exposure to

new lexical items in different contexts, e.g. the document retrieval on the Web (Grabe and Stoller, 1997).

Other researches confirm that there is a meaningful learning of new terms, above all when there are connections

between new words and others whose meaning and context of use are already known. Therefore, the conceptual

maps approach can ease the understanding of new terms and of their relationship to the knowledge previously

acquired. It favors a multiple cognitive exposure to the lexical elements and encourages the connections among

terms belonging to different contexts. The results issued by search engines often show how a concept/word can

appear in several different knowledge domains (contexts) in a transversal way, as theorized by Foucault and

Guattari: to recognize this transversality means to be able to connect apparently separated domains in order to

build alternative conceptual representations, useful to look at a problem with new eyes.

Example of an application of the SEWCOM method:

how to use the “cognitive dissonance” of search engines as a creative resource

Our search may be Galileo Galilei, the famous scientist of 1600. Our goal is to limit the hits concerning not

pertinent documents, and to find an appropriate strategy using some metacognitive considerations. We notice a

strong effect of cognitive dissonance (Festinger, 1978), probably because of the surprise to find documents with

different semantic meanings with respect to the search topic.

In some cases the dissonance was easy to solve: for example, for some documents related to the NASA Galileo

project we learnt that the reason of its name was that it was a probe sent to study Jupiter and its satellites, that

had been discovered by Galileo. In other cases the dissonance was persistent and sometimes it was not noticed at

all at a conscious level, so the document was immediately rejected without an additional attempt to investigate

further. In order to limit the effects of the cognitive dissonance, we try to reformulate the search starting from a

"brainstorming", and using the knowledge we have of the topic (Jonassen and Gabrowski, 1993). So we began to

associate to the "Galileo" concept as many meaningful words as possible, and to visualize their relationship by

means of a concept map. Sometimes single words were proposed, such as "telescope" or "abjure"; sometimes

groups of words like "Copernican revolution", "force of gravity" or entire sentences that described a concept

entirely, as "The world turns around the sun", or "the creator of the scientific method".

The words/concepts obtained by brainstorming had to be re-organized into the map space. It was thus possible to

create some semantic areas, from which we could select the terms to be included in the search query. This way,

each argument could be focused better. For example, the semantic area "religion" or Galileo's relation with the

Roman Church came out often (keywords: "Inquisition", "Pope", "Holy Scriptures"); so did the area related to

"astronomy" (keywords: "Sun", "Earth", "Jupiter", "Telescope").

The borders limiting the region including the words can be redrawn in a different way, changing their shapes and

colors so as to improve the visual appreciation of the map and the perception of its different areas. We realized a

kind of visual categorization. Besides the usual list of hits, some search engines like Teoma or Vivisimo try also

to offer the user a list of automatic categories in which the results can grouped. Anyway, this kind of automatic

classification is not really satisfactory, because the user cannot choose the categories autonomously. The

collaborative effort in the brainstorming step led to focus a larger set of words, where more specific keywords

could be selected in order to refine the search process.

When searching on line, quite often some problems are encountered that stem from homography, polysemy or

synonymy. Even though two persons share the same concept, it is highly unlikely that they express it with the

same words: "the probability that two persons use the same terms describing the same thing is less than 20%”

(Bates, 1986).

Successive searches based on the terms included in the different semantic areas were capable to limit the number

of the hits. For example, we obtained about 44,100,000 references with the term "Galileo", about 323,000 with

"Galileo and Inquisition", about 21,700 with "Galileo and Abjure", but only 793 with “Galileo AND Inquisition

AND Abjure AND "Pope John Paul II". According to this, during the second phase of the search we could notice

a strong relation between the knowledge of a specific language of the domain under investigation and the ability

of "reducing" the search on-line. In fact the term "abjure" didn't appear frequently, as it is a word seldom used in

everyday language, and strongly bounded to particular contexts.

At the final step, pictures were inserted near the nodes in order to make the map more appealing and to stress the

"discovery" of new terms/concepts. For examples those of the "Foucault's pendulum" (fig. 3) or those on the

"perfection of the celestial bodies" (dogma of the Roman Church questioned through the use of the telescope).

These terms/concepts were identified in the documents retrieved via the search engines. When we thought they

were important, we inserted a link to them in the map.

The new terms/concepts must now "justify" their presence with a creative re-organization of the map that often

“produces” new knowledge. It is interesting to notice that this new knowledge is transversal with respect to

specific domains. It is also important to notice how new ideas may stem from the interaction of different

domains: for example, the demonstration of the rotation of the Earth using only a pendulum instead of optical

observation, or the importance of the relation between science and religion in Western History.

We could thus realize how what is usually considered as a drawback can be converted into a precious resource:

knowledge is transversal, and this favors an educational planning based on multidisciplinary approaches.

References: http://multifad.formazione.unipd.it/sewcom