The planned undertaking is to plan and make an Expert Artificial Intelligence System that is capable of analyzing natural English text, which a user inputs, to place the user ‘s specific auto demands. Once they have been identified the system will react really smartly with a list of autos that matches the user ‘s standards.

I will be making this expert system in C++ utilizing Natural Language Processing and Artificial Intelligence with a back-end database, where the information of autos will be stored. The manner in which this system works is straightforward for the end-user. The terminal user will be able to come in a few lines of Natural English Text into the system, such as ‘I am looking for a black BMW with a 2.0 liter engine and it has to be a manual auto which had done less than 50,000 stat mis ‘ . The system will now pull out the of import pieces of informations like ‘black ‘ ‘BMW ‘ ‘2.0 ‘ ‘manual ‘ ‘less than 50,000 ‘ and set them into a tabular array and run a question to the database to happen any duplicate autos. Finally the system will react with a list of available autos that matches the user ‘s standards.

Research shows that presently there is no such system of this sort available worldwide and this will be one of the first plans being made for this kind of market. This type of plan is really alone as it will assist seek for autos with merely a few lines of English text, alternatively of traveling through 15 or 20 boxes make fulling out your auto demands. This plan should turn out utile to those who are seeking to seek for autos with more easiness and will supply a more specific and greater set of consequences for the terminal user.

Once a working beta of this plan is made, it will be tested with the aid of the general populace interested in autos as they are the aimed market for this plan. With the consequences of the testing, the plan will so be implemented to suit the needed maps.

Background

In this coevals about everybody who is legal to drive owns a auto or is in hunt of a auto. From the age of 17, when childs are allowed to take control of the wheel, they are in hunt of their first auto. On mean each house clasp in the UK owns at least 2 autos. From clip to clip people prefer alteration and it ‘s clip to get down looking for a new auto, and the demand for autos will ever stay. The bulk of the population nowadays prefer to purchase 2nd manus autos as they are better for value. Besides due to the recession most people at the minute can non afford to purchase a trade name new auto but alternatively prefer to purchase a 2nd manus vehicle. Peoples will utilize resources on-line and local resources to seek for their new auto, but sometimes struggle to happen their ideal auto. Possibly they are in demand of an easier on-line hunt engine to happen the auto most suited for them.

“ The used auto market is at the minute a monolithic market, because of the recession there is a batch of people looking to purchase a inexpensive 2nd manus auto. The recession in old old ages made the used auto industry a monolithic sum of money. ” [ 1 ]

Natural Language Processing

To do this plan achievable the system should recognize Natural English Text and be able to pull out the keywords. Imagine when you ‘re face to confront with a gross revenues representative, and you tell him what type of auto you are interested in, he would hold to understand to the full what precisely your demands are and in the same manner the system should be able to understand what precisely the terminal user is inquiring for.

“ The end of the Natural Language Processing ( NLP ) group is to plan and construct package that will analyse, understand, and generate linguistic communications that worlds use of course, so that finally you will be able to turn to your computing machine as though you were turn toing another individual. This end is non easy to make. “ Understanding ” linguistic communication means, among other things, cognizing what concepts a word or phrase bases for and cognizing how to associate those constructs together in a meaningful manner. It ‘s dry that natural linguistic communication, the symbol system that is easiest for worlds to larn and utilize, is hardest for a computing machine to get the hang. Long after machines have proven capable of inverting big matrices with velocity and grace, they still fail to get the hang the rudimentss of our spoken and written linguistic communications. ” [ 2 ]

Artificial Intelligence

Artificial Intelligence ( AI ) is normally defined as the scientific discipline of doing computing machines do things that require intelligence when done by worlds. There are several types of AI methods I could utilize to transport out this undertaking but I have chosen to travel in front with utilizing the regulation based method and besides a spot more complicated method, the fuzzed logic method.

“ Q. What is unreal intelligence? A. It is the scientific discipline and technology of doing intelligent machines, particularly intelligent computing machine plans. It is related to the similar undertaking of utilizing computing machines to understand human intelligence, but AI does non hold to restrict itself to methods that are biologically discernible. Q. Yes, but what is intelligence? A. Intelligence is the computational portion of the ability to accomplish ends in the universe. Changing sorts and grades of intelligence occur in people, many animate beings and some machines. ” [ 3 ]

Wider Context

In a wider context this piece of package can be used by any company in the universe who sell used autos. It can be an excess characteristic to their web site or even a plan they can utilize on a day-to-day footing to seek for autos for their ain clients.

2.0 Aim

Specific

To plan a plan created in C++ , which is capable of utilizing Natural Language Processing and Artificial Intelligence with a backend database, which will interact with you and assist the terminal user select a suited auto.

Measurable

The plan has to accept Natural Language to be able to separate what the user is inquiring for. Besides the system has to be able to utilize Artificial Intelligence to a question a database to acquire the consequences and so be able to show accurate consequences from the database in a manner which is clear and apprehensible by the user.

Accomplishable

I have yet to find the best manner of utilizing Artificial Intelligence to show accurate consequences from the database back to the user, so research in this country will be on traveling. Besides an ample sum of proving of the system will necessitate to be carried out to do certain the system is in working order.

Realistic

A Gantt chart has been created that will be followed, this will do my ends accomplishable and the coveted consequences should hence be achieved.

Time

There are several deadline preset that will be followed, reexamine point one, reexamine point two, the concluding deadline and presentation. Equally good as these there will are deadlines set for my accomplishments every two hebdomads, and a meeting with my Undertaking Tutor. These are all shown on the Gantt confab that has been created.

3.0

Project Scope

Within Scope

A plan that can place the user ‘s demands from natural English text and be able to propose positive consequences to the user, showings them a list of autos that match their demands. It may besides be able to demo alternate consequences if there are no autos fiting their standards.

Out of Scope

A plan that can work accurately with every type of text the user inputs and be able to react with accurate consequences every clip. It will be hard for the plan to understand the user ‘s demands if the text has n’t been typed accurately. For case if the user enters ‘blak ‘ alternatively of ‘black ‘ , the system will non be able to recognize the word and hence will non be able to show accurate consequences. The plan will ne’er be able to understand any brief words that have been typed into the system. Once people start utilizing the cyberspace more and more and get down utilizing chat applications such as MSN, they tend to bury how to spell in full English and utilize short manus footings for words such as ‘u ‘ alternatively of ‘you ‘ and there are plentifulness of words that can be cut down to rush up their typewriting. This is a wont people catch onto while typing and could do some issues when utilizing this plan.

Milestones and Deliverables

Milestone 1

Research the methods, techniques and characteristics for a successful AI Expert System

Transporting out primary research, utilizing books & A ; the cyberspace.

Transporting out secondary research.

Research the facets of the undertaking

Transporting out secondary research – e.g. how to utilize AI to acquire the plan working, how to incorporate a database into the system

Research how the plan will work.

Deliverable

A elaborate specification of the undertaking should be produced and documented and the Project Planning Document for Review Point 1 should be prepared.

Milestone 2

Transporting frontward from Milestone 1, develop possible solutions to run into the demands, utilizing the cognition gained from research.

Deliverable

Document possible thoughts, with the concluding pick of how the undertaking will be implemented.

Milestone 3

Produce initial designs of the paradigm, listing required functionality collected from questioning possible users.

Make a elaborate trial program, which will be used one time the application has been completed. The trial program should include, user testing, development testing, integrated testing, functional testing, public presentation testing and unit testing.

Deliverable

Create initial designs of the paradigm and papers a trial scheme. Hand in for Review Point 2.

Milestone 4

Assuming the paradigm maps right, the system can go on to be developed.

All proving internally and externally should be carried out, guaranting all maps and characteristics are working right.

Deliverable

The paradigm execution should be complete, with certification.

Trial Plan should be completed and documented. Assuming clip licenses, solutions of any mistakes in proving should be resolved.

Milestone 5

Assuming the merchandise is complete so if the clip permits, the concluding phase is to implement the application in a practical environment.

Deliverable

If the merchandise has been completed, and there is clip available, the application should be implemented into a practical environment, with certification.

The option is certification should be produced ; placing clip restrictions and future work on the undertaking.

Milestone 6

Finally, the thesis should be written up, checked and handed in by 28th April 2010

The undertaking presentation should be carried out on 29th April 2010

Deliverable

Undertaking Thesis

Undertaking Demonstration

4.0

Beginnings of Information

For this undertaking the beginnings of information will change. Here are some illustrations of beginnings of information that might be used to run into the range of the undertaking:

Internet – immense sum of resources for every phase of the undertaking

University – class notes from first twelvemonth and 2nd twelvemonth talks.

Library – resources of books

Literature – a assortment of books and publications on AI, Expert System, NLP, GUI Interfaces, C++ , information protection act, advertisement and concern direction.

General Public – can be used as a resource for inquiring inquiries about the demands of autos.

Paul Bowden – Undertaking coach to steer and back up throughout the undertaking.

Here is a list of beginnings of information required to carry on the undertaking:

Microsoft Office 2003/2007 – Documentation of the undertaking

Microsoft Project – to bring forth a Gantt chart

Microsoft Visual Studio – for package building

Microsoft SQL Server 2005 – to make SQL databases and shop informations here.

GWIN – to make a GUI interface

5.0

Undertaking Risks

With any undertaking that is undertaken, there are many hazards that will or may happen. In order to be prepared for these hazards to happen, a eventuality program has been put in topographic point to forestall catastrophe taking topographic point.

The Hazard analysis is being done with the usage of a hazard tabular array, this will assist to set each hazard on a graduated table and dressed ore on the most sever hazards.

The tabular array is laid out into 5 columns: Hazard, Probability, Severity, Total Risks and Counter Measures. Both chance and badness are scaled from 1 to 3. Both consequences for each Risk point are multiplied to bring forth the Total Risk. The Entire Hazard should so be compared to the Key ( see below ) , therefore enabling more attending to be given to the most sever hazards.

Key

Entire = Severity x Probability

Badness

Rating

Probability

Rating

Entire

Rating

High

3

High

3

6 – 9

High

Medium

2

Medium

2

3 – 4

Medium

Low

1

Low

1

1 – 2

Low

Hazard Table

Hazard

Probability

Badness

Entire

Counter Measures

Time: Not being able to finish the undertaking on clip, due to clip limitations.

2

2

4

Reviewing the Gantt Chart on a regular basis and prioritizing.

Data Loss: System failure taking to loss of work.

1

3

3

Regular backups should be made in multiple locations.

Lack of resources: if cardinal hardware and package unavailable to finish the undertaking.

1

2

2

Guarantee all hardware and package entree is available, in progress.

Scheduling: If there are troubles with the scheduling or cognition spreads.

1

2

2

Keep researching and reading books to derive more cognition in the capable country.

Illness: If for any ground I fall badly during the clip of this class.

1

2

2

Attempt to finish undertakings before deadlines, leting excess clip.

Discuss with coach what can be done for the lost clip.

The most serious hazard is merely rated as Medium, and that is, non following the Gantt chart and mileposts. If these are non followed so the undertaking will acquire delayed, and this will finally take to me working through the concluding phases of the undertaking at the last minute and even do the undertaking to be uncomplete by the deadline.

If the deadlines are non met, longer working periods will hold to be started. For me to avoid this job a hebdomadal agenda has been made for me to follow.

The remainder of the hazards are acceptable and ineluctable. I will merely follow the counter steps so if they occur readyings have already been made.

6.0 Evaluation of Professional, Social, Ethical and Legal Issues

With a merchandise of this nature, there are many professional issues, affecting legal, societal and ethical facets of information systems. The major considerations are outlined below:

Confidentiality – The personal information of the auto proprietor will be stored and used to supply them with promotional stuff ; this will necessitate consideration of the Data Protection Act ( 1984, 1998 ) to protect their informations.

For protection and encoding of client inside informations, the Computer Misuse Act 1990 will besides be relevant.

Copyright and Ownership – Due to the nature of this undertaking, particular attending will be paid to the promotional stuff produces particularly where literature and cryptography is involved. All beginnings of information must be checked for right of first publication and ownership if they are to be used in the promotional stuff.

7.0 Gantt Chart