Using Digital Dictionaries and Thesauri to Improve Predictive Ajax Search Interfaces

1. Project Definition

This project is intended to investigate the use of predictive and type-ahead text in the context of search engine interfaces. It will examine how we can improve web-based search interfaces using controlled thesauri and dictionary lexical and semantic resources, and present the information in a timely fashion using Ajax and Web-Services.

The scope of this project could potentially cause problems because the area of Ajax type-ahead search interfaces is relatively immature. Consequently, the continually developing and changing nature of the application area could cause the project to inadvertently enter additional research.

It is vital that the aims and objectives pin-point the exact elements to be investigated so as to prevent scope-creep.

1.1 Aims

Define a light-weight Web-Service schema and build a prototype that accesses dictionary and thesauri data. The schema must provide an API (Application programming Interface) suited for use by JavaScript search interfaces that provide users with suggestions using type-ahead mechanisms to suggest how they could refine their search criteria.

As we have already stated, the project is located within a relatively immature application area. The project can be considered to be part of a wider aim: the continuing development of search interfaces and engines to create the ultimate search mechanism that guarantees relevant, quality search results.

In truth this is a "holy grail". Whilst we cannot necessarily expect to achieve this ultimate goal, we can strive to come as close as is realistically possible. By light-weight, we mean the Web-Service must use uncluttered formats and call signatures. For the Web-Service to be fit for purpose it must be very efficient, if it is not, the interface will be unresponsive and the benefits that we are trying to obtain will become diluted.

1.2 Objectives

The objectives are split into four phases, research, design, implementation and evaluation.

1.2.1 Research

Research ways in which digital dictionary and thesauri are accessed and used to improve user interfaces, also consider the use-case scenarios that are present and how they influence dictionary and thesauri APIs.

There are numerous ways in which thesauri and dictionaries can be accessed. We are specifically interested in Web-Services that enable access to these resources. Although Web-Services generally conjure ideas of large XML (eXtensible Markup Language) documents and DTDs (Document Type Definition), a Web-Service can use any format. The project needs to identify, extend or define a format that is suitable for this particular scenario.

1.2.2 Design

Design a Web-Service schema that enables the use-cases, identified in the research analysis, to access dictionary and thesauri data for auto-completion.

Based on the findings of the research, the design phase will generate three main elements: the Web-Service schema which will include the call signatures and data structures; the internal operation of a server prototype; the internal operation of a client that demonstrates the functionality of the server prototype and the Web-Service schema.

The most important element of the design phase is the creation of the Web-Service schema. The server and client prototypes are only intended to demonstrate the use of the schema and the suitability of the schema to the problem area.

Depending on the number and complexity of the use-case scenarios identified in the analysis of the research, it may be necessary to restrict the Web-Service schema to a small number of these.

At this stage it is assumed that the client prototype will use type-ahead techniques to tailor the user interface. Great care needs to be taken when evaluating the effects of type-ahead on the client lag and the server load.

1.2.3 Implementation

Implement a server prototype that adheres to the Web-Service schema and an Ajax enabled client prototype that demonstrates the use of the server.

The server and client prototypes need to be implemented at this stage as proof of concept. The rationale behind the prototypes is solely as a source for the evaluation; they are not intended to be used in a live system.

1.2.4 Evaluation

Evaluate the overall responsiveness of the system and discuss the future of the technology, identifying areas that need to be improved.

This project is a WIP (work-in-progress), although the evaluation will attempt to draw concise conclusions, it is likely that the conclusions will be built on and scrutinised further in the future.

The question which we really want to answer is "At this moment in time, does the defined Web-Service schema make for a suitable system on which to build a search-interface that uses type-ahead techniques to display suggestions sourced from a managed lexical data source?".

1.3 Deliverables

The project will be delivered in three milestones, this is the first.

Milestone 1. Project Definition and Research
Completion Date: 21st Nov 2007

Milestone 2. Research Analysis and Prototype Design
Completion Date: 30th January 2008

Milestone 3. Prototype Implementation and Evaluation/Project Evaluation and Conclusions
Completion Date: 16th April 2008

The implementation of the prototype will start before the end of Milestone 2, but finish during Milestone 3.

At the end of the project there will also be the opportunity to attend a presentation and questions and answers session which will attempt to summarise the project and the findings and conclusions drawn from it.