CONTEXT-AWARE SEARCH ARCHITECTURE

Hadas Weinberger, Oleg Guzikov and Keren Raby

HIT - Holon Institute of Technology, Holon, Israel

Keywords: Adaptive Systems, Context, HCI, Search Engine, Recommender System, User Model, Web 2.0, Semantic

Web.

Abstract: There are several reasons for developing a context-aware search interface. In so far, search engines

considered the technology perspective – suggesting structural, statistical, syntactical and semantic measures.

What is yet missing in Web search processes is the inclusion of the user model. The prevailing situation is a

usability hurdle.

While there is a wealth of information about search engines, what is yet lacking is a recommender system.

Such as could be provided by a set of adequate principles and techniques, as basis for the design of a Web-

base interface guiding users towards efficient and effective utilization of the spectrum of search engines

available on the Web. The research reported here takes a step towards this goal, suggesting context-aware

search architecture (namely, CASA) aiming towards: 1) the analysis of query elements, 2) guiding the

process of query modification, and 3) recommending the personalized use of search engines.

A use case illustrates the need for the suggested framework and a prototype Web interface is introduced. We

discuss preliminary findings from empirical research conducted with several classes of students in two

distinct academic institutes in two different countries, which concerns the feasibility and usefulness of the

suggested framework. We conclude with recommendations for further research.

1 INTRODUCTION

The interface level of a Web search process involves

three elements: 1) the user’s query, 2) a search

engine and 3) the search results. Two out of these

three elements are anchored in user’s context. First

is the user’s query, which is often subject to

negotiation and modification. The query represents

the user’s model (Marchionini and White, 2007) as it

is established by the context of the investigation

(Marchionini, 2006). Second are the search results

that should respond to the query and reflect its

context. Search engines, however, are usually

approached independent of the user’s context

(Kritiquo, 2007; Weinberger, 2009).

This situation, albeit prevailing, is disregarding

the opportunities available by search engines’

technology which could be proved useful, enhance

precision and promote utility for the user – provided

they are used in context. This is specifically true for

users engaged in exploratory search – either as part

of business processes or in academic setting

(Marchinini, 2006; White and Roth, 2009; White,

Kules and Bederson, 2005). The prevailing situation,

in which the interface does not allow the selective

use of search engines, is a usability hurdle.

What is yet missing is an interface instructing the

manipulation between search engines in a manner

that considers the user’s model – allowing the user a

choice between different search engines. For this

end the envisioned interface should include tools for

user’s requirements’ elicitation on the one hand, and

for the modelling of the user’s query within its

context, on the other hand.

In order for users to exploit Web search

technology, there is a need for tools and techniques

that would instruct context-aware utilization of

search engines (Vossen and Hagemann, 2007;

Weinberger, 2009). Different than the dynamic and

active role of the user in the Web 2.0 arena, search

interactions remained aloof of the user’s individual

context. In view of the wide spectrum of search

engines (SEs) available on the Web (e.g., popularity-

based SEs, social SEs, semantic SEs, hybrid SEs,

domain specific SEs) it is surprising that there is no

interface instructing search engines’ context-aware

methodological utilization in a manner that

considers the user’s query as part of the user’s

context and with relation to the user’s model.

Weinberger H., Guzikov O. and Raby K. (2010).

CONTEXT-AWARE SEARCH ARCHITECTURE.

In Proceedings of the 12th International Conference on Enterprise Information Systems - Human-Computer Interaction, pages 71-78

DOI: 10.5220/0002968300710078

 SciTePress

While there is much research effort aiming to bridge

the gap between search engines’ methods and the

user’s model (Martzoukou, 2004; Mammr,

ALKhatib, Mostefaoui, Lahkim and Mansoor,

2004), the research reported here takes a slightly

different perspective; aiming to bridge the gap

between the user’s query and the appropriate search

engine focusing on the user’s perspective, i.e., her

context. This research concerns the design of a

Context-Aware Search Architecture (namely,

CASA) to support a search interface that would

facilitate a) an interaction with the user based on b)

the user’s modified query and c) a search engine

recommender system.

Our approach to the design of CASA follows the

design science paradigm (Havner, March, Park and

Ram, 2004; March and Smith, 1995). Of the

research activities outlined by design-science

research in IS this paper covers the build (a Web-

based recommendation system as part of a search

interface) while for the evaluation of this artefact we

report on preliminary (qualitative) results of

empirical investigation. Of the four design artefacts

(i.e., constructs, models, methods, and

instantiations), outlined in these frameworks, this

research is about a model (i.e., the method

instructing the recommender system’s principles),

which informs a methodology (i.e., the techniques

for supporting user’s requirements elicitation and

query modification processes) and an instantiation (a

prototype of the Web interface).

Following this introduction, section 2 holds a

brief discussion of search engines. Section 3

describes the need for context-aware search

architecture and section 4 describes this architecture,

i.e., CASA. Section 5 is focused on the methodology

used in this research. We conclude in section 6 with

a summary and discussion.

2 SEARCH ENGINES IN

CONTEXT

The lack of a consistent methodological approach to

Web information seeking research (Baeza-Yates,

2003; Martzoukou, 2004) might be attributed to the

dynamic nature of the field. Frequent innovations in

search engines’ technology modify search engines’

classification. Consequently, best practices of the

field are often altered (Vossen and Hagemann,

2007). Currently there are several leading practices

in search engines technology of which we mention

several examples: a) popularity-based SEs (e.g.,

Google) which also manipulate a host of other

algorithms (e.g., statistical measures, Web-genre

analysis, clustering and categorization), b) Inclusive-

meta SEs (e.g., Myriad, Quintura), c) social SEs that

focus on user’s contribution (Hakia, FreeBase), d)

Semantic Web SEs (e.g., Hakia) and analytic SEs

(e.g., WolframAlpha). Other navigation and

information retrieval methods follow notions of:

Web-genre (e.g., Google scholar), domain (i.e.,

geospatial), structure (e.g., Wikipedia) or

phenomenon such as the long tail of search (e.g.,

FeedMil).

Taking the HCI perspective, several SEs include

features that support user’s interaction with the

results as obtained, through activities such as

providing feedback or by allowing navigation and

negotiation of results based on data visualization.

Examples are navigation of interactive maps (e.g.,

Kartoo), user voting (FeedMil), clusters negotiation

and categorization (Clusty).

With the advancement of Web 3.0, there are

indeed innovative technologies embedded in search

technologies (Berners-Lee, Hendler and Lassila,

2001; Finin and Ding, 2006; Ding, Pan, Finin, Joshi,

Peng and Kolari, 2005) that assist in incorporating

user’s annotation (Bao, Wu, Fei, Xue, Su and Yu,

2007) also for the purpose of instructing the user

model (Carmagnola, Cena, Cortassa and Gena,

2007).

However, by the most part users are captivated

by what could be named: ‘the ease of search’

syndrome which prevents them from using multiple

search engines and the options they suggest. As

much as HCI research should approach current

practices (Hochheiser and Lazar, 2007) search

engines’ technology should advance beyond current

context building methods such as: a) structural

attributes, b) syntactical features, and c) semantic

analysis, towards the user’s context (Dey, 2001;

Kobsa, 2001; Midwinter, 2007; Shen, Tan and Zhai,

2005) in order to reflect on the user’s perspective.

For this end, users’ ought to be considered as actors,

allowing them more freedom of action and choice.

Along this lane we mention that classic criteria

for information retrieval evaluation are precision and

recall. While the prevailing practices will not

necessarily promote precision, user’s enhanced

involvement should not be underestimated as an

agent of precision. Against this background a

method and a mechanism could be considered,

which responds to the bi-dimensional view of the

search operation, to include: a) search engines’

typology on the one hand, and b) user’s query and its

context on the other hand. This way search activities

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

would facilitate an efficient and effective search – in

the context of the users’ model as represented by her

query.

3 THE NEED FOR

CONTEXTAWARE SEARCH

ARCHITECTURE

This section brings forward the issue of the user

perspective motivating this research. A concise

discussion paves way to the introduction of a

hypothesis concerning the user’s perspective. We

conclude this section with a use case illustrating the

need for context-aware search architecture and

explain the relationship between the three elements

that constitute the search interaction: the user, the

query and the search engine.

3.1 The User Perspective

Of the two most common tasks which best represent

Web HCI, users’ contribution to online communities

is mentioned alongside search processes. For the

former, Preece and Shneiderman (2009) identified

several distinct types of users participating in online

communities. Their work illustrates a typology that

is based on the classification of user’s contribution –

based on users’ task and role, identifying three user

types: a reader, a contributor and a leader. Similar to

this user’s classification, search processes are

classified by three search types: a simple search,

learning and investigation (Marchionnini, 2006)

based on a task-related perspective. Specifically, the

tasks considered for user’s classification differ by

the value assigned to attributes such as the frequency

of the iteration, the complexity of the issue at hand,

and the context.

In the context of these two typologies, this

research assumes value for the contributor and for

the leader who are engaged in either learning or

investigation. A search that is conducted in this

context is likely to motivate users towards the

cultivation of adapted search habits that would yield

useful results. This assumption might be specifically

true in the case of experienced users (i.e.,

contributors or leaders). This brings us to suggest the

research hypothesis that concerns the user

perspective.

Hypothesis 1: Search operations are mostly

conducted using popular search engines

while user’s navigation between search

engines is not a common phenomenon. This

is not because search engines are all alike,

nor is it because different search engines

would not yield different results, but because

adequate context-aware recommendation,

personalization and adaptation tools for this

end are yet missing. Given the adequate

tools, users’ search behaviour might be

altered. One possible path would be for the

search domain to develop similarities with

Web 2.0 tools – where diversions between

knowledge sharing tools and online

communities are not only acknowledged but

are also integrated in services suggested for

users and by users’ practices.

3.2 An Example use Case

A user involved in an exploratory search session is

facing two challenges that concern the ‘how’ and the

‘where’ of the search process: 1) how to search

refers to several activities related to syntactic and

semantic search features such as: choosing key

words, query structuring, modification and the

identification of the domain and the genre to be

explored, 2) where to search is about which search

engine to use. While the latter might appear to be a

decision motivated by the technological perspective,

there are other relevant perspectives to be included.

For instance, we mention the SE’s scope and HCI

features.

As part of selecting a search engine, the user is

required to meet challenges that concern his interests

as well as challenges belonging with Web

proficiency. While there are a host of interface-

embedded syntactical, structural, semantic and

statistics features that would support query

formulation, there are no interface features directing

search engine manipulation. In view of the spectrum

of search technologies available on the Web, there is

not only much promise that is yet unexploited, but

also a serious challenge summoned for the user.

While users might occasionally be aware of the

plethora of search technologies, they still need a

good reason to use these tools. For instance, a social

search engine might yield different results than a

popularity-based search engine, since each uses

different tracking and indexing methods. As an

example, consider the case of a user seeking

information about blogging, more specifically: 'how

to write a successful blog'. While a popularity-based

search engine is likely to return the most popular

results, a semantic search engine such as Hakia is

likely to return results that origin with user’s (recent)

input, indicating an innovative guide, tool or

CONTEXT-AWARE SEARCH ARCHITECTURE

practice. This is not to say that one result is

preferable to the other – but to put forward the

differences that prevail. In the context of the design

of business, research or learning environments, we

would like the user – a reader or a contributor, to be

aware of her options in a most profound way.

There is in this use case to a) demonstrate the

need for context-aware search architecture, b) to

describe the relationships between the three

elements that are part of a Web-search interaction,

and to c) anchor the former two as part of a wider

perspective on current HCI challenges.

4 CASA: CONTEXT-AWARE

SEARCH ARCHITECTURE

The CASA architecture is comprised of a 1) two-

faceted query definition and modification

mechanism, and a 2) set of recommendation

principles guiding the process of search engine

selection. The OSKA-based (Weinberger, 2010)

search interface (Web: http://oska-search.info/) is a

prototype demonstrating the operation of the

framework suggested here. This prototype provides

users with an example experience – albeit not fully

supported, for a search interaction that utilizes the

method presented in this research.

4.1 The Search Interface

A Web-based system demonstrates the method of

the CASA-based recommender system. The

interface is designed to respond to a) the user’s

query by suggesting an adequate use of b) a search

engine.

For the design of the user interaction (building

the query’s context) we follow the Ontology for

Social Knowledge Applications, namely, OSKA

(Weinberger, 2010) intended for aiding users

throughout the annotation of Web 2.0-tools user-

generated content and context. Since tagging and

search are considered the two sides of a coin (White

et al., 2005), we assume the ontological construct

could be followed for user’s requirements elicitation

and for query modelling.

The prototype interface (Figure 1) demonstrates

the support available for the user in determining the

a) query’s current focus (i.e., there, Query type) and

choosing b) an ontological extension (i.e., there,

Question type). Based on this ontological analysis

the system c) recommends the search engine that is

likely to yield results that are of highest precision –

in accordance with the recommendation principles

(described herein).

The mechanism for identifying the Query type

responds to the three-perspective view identified for

the Ontology for Social Knowledge Applications

(i.e., content, task and technology). The mechanism

for the identification of the question type follows the

WH questions scheme used also in the IS field for

the evaluation of information systems.

4.2 Recommendation Principles

This section is dedicated to the five recommendation

principles identified for this research. The

description of each recommendation principle (RP)

is anchored in the context of a search engine type in

relation to the WH question (i.e., aspect) for which it

best responds. For each search engine type we

provide example evidence description, annotated by

a-e, followed by a Recommendation Principle,

formatted with bullets.

A. Popularity-based Search Engine – e.g., Google:

the results tend to spread across several aspects

of the query element(s); Answering questions

such as: what, hence facilitating an introduction

to the subject domain. Based on this finding the

following RP was formulated:

• RP1: A search for general information that

is spreading across several aspects (i.e.,

responding to WH questions such as ‘what’), is

likely to be useful by means of using a

popularity-based search engine, e.g., Google.

B. A social-semantic search engine, e.g., Hakia: the

results tend to focus upon example instances of

the query’s element(s); Answering questions

such as: what, how and where, hence enabling

the study of example applications. Based on this

finding the following RP was formulated:

• RP2: A search for information describing

attributes assigned to a certain concept (i.e.,

responding to WH questions such as ‘what’ and

‘how’), is likely to be useful by means of using a

social-semantic search engine such as Hakia.

C. A Semi-semantic and Visualized Search Engine,

e.g., Kartoo, Clusty: the results tend to spread

across three instance-level aspects, answering

questions such as: who, how and where;

facilitating the comprehension of a phenomenon.

Based on this finding the following RP was

formulated:

• RP3: A search for instance-level responses

to questions with relation to a specific domain

(i.e., responding to WH questions such as ‘who’,

‘how’ and ‘where’), is likely to be useful by

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

means of using a semi-semantic, clustering,

visual or interactively enabled search engine

such as Kartoo or Clusty.

D. An Analytic, Semantic-, Social-semantic or

Hybrid Search Engine, e.g., FreeBase, FeedMil:

the results tend to focus on several practical

aspects, answering questions such as: where and

how, responding to the technology perspective,

hence summoning the user a wealth of

information allocated by users to guide the

investigation of a subject domain. Based on this

finding the following RP was formulated:

• RP4: A search for instance-level

information based on user-input (i.e., responding

to WH questions such as ‘where’ and ‘how’), is

likely to prove useful by means of using an

analytic, semantic- and social-semantic or hybrid

search engine such as FreeBase or Feedmil.

E. An Analytic Search Engine, e.g., WolframAlpha:

the results tend to focus on several aspects,

presented as a report on the subject of the

investigation – based on a dialogue with the user.

Specifically this search engine will prove useful

for the user facing a depth- and wide-motivated

search. Based on this finding the following RP

was formulated:

• RP5: A search for a wide perspective

perception of a domain is likely to be found

useful by means of using an analytic, semantic

and hybrid search engine such as WolframAlpha

that is empowered by artificial intelligence –

amongst other features. This search engine

compiles a categorized report of the subject

matter, not only introducing the user a host of

information in various forms but also allowing

him the negotiation and analysis of the

presentation of the findings.

4.3 Instructing a User-cantered Search

In this section we illustrate an example use case of

utilizing a user search interaction by the method and

principles prescribe in CASA using the OSKA-

search (Web: http://oska-search.info/) prototype

(Figure 1) aforementioned. There are three stages in

this interaction. For each stage, the user’s role and

the system’s response are described.

Stage 1 – Search Initialization: user introduces

query elements in the search box. For example, the

query element may be the expression: blog. The

system then identifies the query’s dimension (there,

query type).

Stage 2 – Query Modification: there are two

dimensions to the action lanes defined for this stage.

The first concerns the system perspective and the

second concerns the user’s perspective. The system

perspective prescribes two complementary actions

and decisions, accordingly. The first concerns the

query type and the second concerns the question

type. The first would be feasible provided an

adequate lexical ontology is available. That would

allow the system the automated identification of the

query type. Second is the identification of the

question type. In this context, the system is designed

to respond to three types – responding to 3

ontological dimensions identified in OSKA

(Weinberger, 2010): subject (i.e., scope), activity

(i.e., task) and media (i.e., technology).

The user’s perspective also involves two actions

lanes and decisions, accordingly. First, the user has

to choose a question type following which the

system suggests to him an extension aspect. For

instance, if the query includes a term such as ‘blog’

that it is identified (i.e., by the system) as ‘subject’;

consequently, a corresponding WH questions (there,

question type) are suggested (Web: http://oska-

search.info/) to further focus the query. For instance,

suggesting the ‘how’ or the ‘where’ extensions. This

procedure is an example for a user’s requirements

elicitation process that is followed by a

corresponding query modification process provided

by the system’s part.

Stage 3 – adapted-personalized search: based on the

previous two stages, the system uses the

recommendation principles mechanism to offer for

the user results wthat origin with the most

appropriate search engine for the query type and in

accordance with the question type.

5 METHODOLOGY

The iterative development of CASA follows the five

stages of system development: planning, analysis,

design, implementation and evaluation. This process

is discussed herein.

Planning & Analysis: involved the consideration

of 1) the search engines to be included in this

research, and of 2) the search terms to be used for

query formulation. Several trial quarries were run

using different search engines for the purpose of

identifying appropriate (i.e., unique and

distinguished) search engines and terms in a manner

that will assure heterogeneity of technology and

ontological diversion of query elements.

CONTEXT-AWARE SEARCH ARCHITECTURE

Figure 1: Context-aware search interface.

For the latter we have found theoretical grounds in

the Ontology for Social Knowledge Application

(OSKA; Weinberger, 2010). Eight search engines

were selected based on the distinct definition of each

and following hands-on, ongoing experience. The

selection process was motivated towards

emphasising the novelty of the search engine

technology, to include: Google as a popularity-

based SE, Hakia as a Semantic- and Social-Semantic

SE, Kartoo and Clusty as visualized and clustering

SEs and FreeBase as analytic and social-semantic

SE. Last but not least are Feedmil and

WolframAlpha. The former is a social-, hybrid and

long tail search engine and the latter is an analytic-

and semantic search engine (see section 2).

Design: was focused upon 1) query formulation

– in accordance with the ontological perspectives of

OSKA. Query elements were defined to meet the

three perspectives view of the aforementioned

ontology. For each search engine three queries were

introduced, using: a) an element of the content

perspective (e.g., Web 2.0 tools, Web 2.0 software,

social media applications, b) an element of the task

perspective (e.g., collaboration, participation,

publishing, editing, reporting) and c) an element of

the technology perspective (e.g., bookmarks, blog,

Wiki, Microbloging, Database).

Yet as part of design we managed 2) the

modification and extension of the list of search

engines, alongside 3) analysis of search results by

the six WH questions. An analysis and

documentation scheme was designed specifically for

this end (Figure 2). This scheme is also used for in-

class assignments as part of students education

towards the implementation of the method suggested

here.

Implementation: involved 1) the definition of the

recommender principles. This was done based on the

analysis of previous results. The analysis and design

process followed the WH questions in order to

identify the relationship between a search engine and

an ontological perspective, using the aforementioned

analysis and documentation scheme. The findings of

the former activities (i.e., the recommendation

principles and the query modification techniques)

were used for 2) the design of the prototype Web

interface. Last but not least we mention integration

into curriculum of the advised method reported in

this paper.

Evaluation: evaluation in this research followed

two lanes. The first is evaluation through design and

the second is empirical evaluation aiming at the

feasibility and usefulness of the architecture – the

RPs and query modification techniques. We

elaborate on the latter evaluation course.

First, the method, as embedded in the design of the

Web interface (prior to design) was introduced as

part of the ‘Web technologies’ course syllabus in

two distinct university classes. The first is graduate

students of a business class in the University of

Nicosia, Cyprus and the second id undergraduate

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

students of our institution. All the students followed

this method for the allocation of resources for their

term projects. These projects were aiming at:

developing subject-specific knowledge sharing

systems (e.g., Bibsonomy, Twine, Google

Bookmarks), the design of Mashup application (e.g.,

iGoogle, netvibes) and of Web-based Learning

Objects – using a spectrum of Web 2.0 tools.

Figure 2: An example analysis and documentation scheme.

6 CONCLUSIONS

This paper suggests a context-aware search

architecture which supports several processes and

activities such as the: a) identification and the

modification of user requirements and the b)

manipulating between search engines, hence,

facilitating the use of a c) recommender system

based on recommendation principles embedded in

this architecture and demonstrated by the prototype

Web interface.

The suggested framework – including the method,

the recommendation principles and the system

architecture, was developed in accordance with the

hypothesis suggested earlier in this paper regarding

usability obstacles in practicing conventions. The

context-aware search architecture advised in this

research utilizes a spectrum of search technologies,

determined based on the relative value predicted for

the user’s model, while aiming towards enhanced

precision.

There are three deliverables to this research in

accordance with the three goals set for this research.

The first is the method, prescribing guidelines for

context-aware query modification. Second are the

recommendation principles directing the utilization

of search engines in context and serving as basis for

the design of the architecture as demonstrated

through the prototype Web interface, which is the

third deliverable that builds on the former two

deliverables to suggest a user- adapted HCI

experience. This interface allows users – regardless

of their domain of practice, in enterprise setting as

part of business interactions, or else as part of

academic setting; an innovative, dynamic and

context-aware search interaction.

The findings of this research indicate a

relationship between a) the search engine type and

the ontological perspectives of the query on the one

hand, and between b) the results obtained by the

search operation, on the other hand. For this reason,

a search engine can be recommended, and the query

may be modified, based on the identification of the

query’s aspect.

There are several limitations to the research

reported here. Indeed, based on our experience the

feasibility and the usefulness of our method were

demonstrated in the field. However, further

empirical evaluation can be carried out to extend

beyond the scope of the examples used here, as well

as with regard to search engines and quantitative

results.

We believe that the findings from our study have

implications beyond this immediate setting. Several

further research directions may be instructed based

on this research. First, we mention the automation of

the interface features, which could be supported

provided adequate ontologies, for instance as part of

Web 3.0, are incorporated as part of this

architecture. Second is the extension of the Web-

interface beyond the prototype features introduced

here. Last but not least is the inclusion of Semantic

Web (i.e., Web 3.0) technologies, such as artificial

intelligence and natural language processing, for the

next-generation of the suggested framework.

All in all, CASA, as suggested here, can improve

and expand the current Web search experience of

individual users, organizations or designers. This

work should prove useful to anyone considering the

development of Web search architecture, or else

individuals seeking to enhance their exploratory

search experience.

ACKNOWLEDGEMENTS

This paper builds on a project conducted under the

supervision of the first author at our department of

Instructional Systems in HIT with two students who

are the co-authors. This paper also builds on my

experiences in recent years with several classes

at our institution and with students of the Business

CONTEXT-AWARE SEARCH ARCHITECTURE

school of the University of Nicosia, Cyprus, were I

was visiting professor on the summer of 2009. I

wish to thank the participating students for their

motivation and cooperation.

REFERENCES

Bao, S., Wu, X., Fei, B., Xue, G., Su, Z., and Yu, Y.

(2007). Optimizing web search using social

annotations. International World Wide Web

conference (IW3C2). WWW 2007, Banf, Alberta

Canada.

Baeza-Yates R. (2003). Information Retrieval in the Web:

beyond current search engines, International Journal

on Approximated Reasoning, 34 (2-3), 97-104.

Carmagnola, F., Cena, F., Cortassa. O., Gena, C. (2007).

Towards a tag-based user model: how can user model

benefit from tags? Lecture notes in computer science,

Springer, Heidelberg, Germany.

Ding, L., Pan, R. Finin, T., Joshi, A., Peng Y., and Kolari,

P., (2005). Finding and Ranking Knowledge on the

Semantic Web, in: proceedings of the forth

International Semantic Web Conference (ISWC'05).

Dey, A. (2001). Understanding and using context.

Personal and Ubiquitous computing, 4, 4-7.

Finin, T., and Ding, L. (2006). Search Engines for

Semantic Web Knowledge, Proceedings of XTech

2006: Building Web 2.0, Amsterdam, 16-19.

Hevner, A., March, S., Park, J., and Ram, S.

(2004).Design science in information systems

research. MIS Quarterly,28(1),75–105.

Hochheiser H., and Lazar, J. (2007). HCI and Societal

issues: A framework for engagement. International

Journal of Human-Computer Interaction, 23(3), 339-

374.

Kobsa, A., (2001). Generic user modeling systems. User

modeling and user adapted interaction, 11, 49-63.

Kritiquo, Y. (2007). User Proﬁle Modeling in the context

of web-based learning management systems. Journal

of networks and computer applications.

Tim Berners-Lee, James Hendler and Ora Lassila (2001).

The Semantic Web. Scientific American, Retrieved

from:

http://www.scientificamerican.com/article.cfm?id=the-

semantic-web

Maamar, Z., ALKhatib, G., Mostefaoui, S.K., Lahkim, M.,

& Mansoor W. (2004). Context-based personalization

of Web services composition and provisioning. Proc.

EUROMICRO, IEEE Computer Society.

March S. T., and Smith, G. F. (1995). Design and natural

science research on information technology. Decision

Support Systems, 15, 251-266.

Marchionini, G. (2006). Exploratory search: from finding

to understanding. Communications of the ACM, 49(4),

41-46.

Marchionini, G., White, R. (2007). Find what you need,

understand what you find. International Journal of

Human-Computer Interaction, 23(3), 205-237.

Martzoukou, K. (2004). A review of Web information

seeking research: considerations of method and foci of

interest. Information Research, 10(2) paper 215.

Retrieved from: http://InformationR.net/ir/10-

2/paper215.html

Midwinter, P. (2007). Is Google a Semantic Search

Engine? Read Write Web:

http://www.readwriteweb.com/archives/is_google_a_s

emantic_search_engine.php

Preece, J. and B. Shneiderman (2009). The Reader-to-

Leader Framework: Motivating Technology-Mediated

Social Participation, AIS Transactions on Human-

Computer Interaction, (1)1, 13-32

Shen, X. Tan, B., Zhai, C. (2005). Implicit user modeling

for personalized search. Proceedings of the 14th ACM

international conference on Information and

knowledge management, 824 – 831.

Vossen, G., and Hagemann, S. (2007). Unleashing Web

2.0: from concepts to creativity. Morgan-Kaufmann.

Weinberger, H. (2009). ECHO: A Layered Model for the

Design of a Context-Aware Learning Experience. In:

Handbook on Web 2.0, 3.0 and X.0: Technologies,

Business, and Social Applications. San Murugesan

(Ed.), Hershey, USA: IGI Global.

Weinberger, H. (2010). Minding Tags for Mining Social

Knowledge (under review).

White, R., Roth, R. (2009). Exploratory Search: Beyond

the Query-response Paradigm. Synthesis Lectures on

Information Concepts, Retrieval & Services. Morgan

& Claypool, Publishers.

White, R.W., Kules B., and Bederson, B. (2005)

Exploratory search interfaces: Categorization,

clustering and beyond. Communication of the ACM.

ICEIS 2010 - 12th International Conference on Enterprise Information Systems