Advert Search Essay (Essay Sample)

INTELLIGENT ADVERT SEARCH BOT AND MARKET RECOMMENDER SYSTEM™
University of Nairobi
School of Computing and Informatics
Maseno C. Khulabe
P15/3889/2008
Supervisor: Dr. Wanjiku Ng’ang’a
Fourth Year Milestone2 Project Report.
© May 2012
Submitted in partial fulfillment of the requirements for the Degree of Bachelor of Science in Computer Science.
Contents
 TOC \o "1-3" \h \z \u  HYPERLINK \l "_Toc321324142"
Abstract  PAGEREF _Toc321324142 \h
ii
 HYPERLINK \l "_Toc321324143"
CHAPTER 1  PAGEREF _Toc321324143 \h
4
 HYPERLINK \l "_Toc321324144"
Back Ground Information  PAGEREF _Toc321324144 \h
4
 HYPERLINK \l "_Toc321324145"
PROBLEM STATEMENT  PAGEREF _Toc321324145 \h
4
 HYPERLINK \l "_Toc321324146"
GOALS AND Objectives:  PAGEREF _Toc321324146 \h
5
 HYPERLINK \l "_Toc321324147"
Project Scope  PAGEREF _Toc321324147 \h
5
 HYPERLINK \l "_Toc321324148"
jUSTIFICATION  PAGEREF _Toc321324148 \h
6
 HYPERLINK \l "_Toc321324149"
Expected Result  PAGEREF _Toc321324149 \h
6
 HYPERLINK \l "_Toc321324150"
CHAPTER 2  PAGEREF _Toc321324150 \h
7
 HYPERLINK \l "_Toc321324151"
LITERATURE REVIEW  PAGEREF _Toc321324151 \h
7
 HYPERLINK \l "_Toc321324152"
Case Studies  PAGEREF _Toc321324152 \h
7
 HYPERLINK \l "_Toc321324153"
History of web advertising  PAGEREF _Toc321324153 \h
7
 HYPERLINK \l "_Toc321324154"
Types And Sizes Of Adverts  PAGEREF _Toc321324154 \h
9
 HYPERLINK \l "_Toc321324155"
Recommender systems  PAGEREF _Toc321324155 \h
10

Abstract
Internet is a generational innovation that has grown extensively over time to take advantage of growing technology, to promote and improve businesses by providing effective communication and efficient service delivery. This has been possible via different forms, mostly advertisements on different on-line sites pages and information repositories on the Internet. However as a result of overwhelming increase of information on Internet, it's difficult to make decisions on where is the appropriate market focus for advertisements or location of products and services. Consequently with advancement in technology and changes in life styles people are preferring different forms of adverts.
The aim of this work is to develop an intelligent advert search bot and market recommender system that crawls through the Internet for adverts, extracts the advert from an online web page into a structured database. The advert is retrieved during search when a matching search term is provided in the search box. This is used to fasten user search and provide more precise advert search needs. This is important since products fall in the vertical internet search domains, where people are interested in specific information rather than the page itself.
This research expounds on prior research that has been done on use of search engines to rank relevant Internet pages; the system is implemented by designing and developing a niche targeting crawler software to extract for Internet adverts. The system allows for external user recommendation on suitable advertising sites to crawl for adverts to simplify customer product and service search and advertisement locations. This simplifies the search since the user does not have to browse through several pages returned by page level search to get the target information.
CHAPTER 1
Back Ground Information
Web advertising is meant to attract potential customers on the advertiser’s Web site of a given product item or to strengthen brand recognition of a business by placing promotional content and a link on other Web sites. Promotional content are designed in various forms to ensure they capture the attention of the target audience. The web has taken advantage of existing technological possibilities of modern Web browsers, to present ads in various forms like: static banners, animated, flashing, moving, links and pop-ups. Most of this became a bother to users where they are not interested in what is being advertised and more irritating if the adverts do not value cultural faith of the user. Research has shown that most advertisements do not serve their intended purpose for instance; according to a report by ADTECH2 the click through rate, i.e. the ratio of the number of clicks to the number of views, has recently fallen below 0.20% on average because most adverts are randomly placed without much consideration of the visitors. With a lot of information flowing on the web it becomes cumbersome to identify relevant advertisements. To solve these problems there is a need to use vertical/focused search mechanisms compared to general page retrieval searches. The vertical search includes Database, machine learning and uses objects as retrieval units as opposed to pages as retrieval units in page level search. It also has powerful query capability, direct answers and aggregate answers. However it has a challenge in where and how to get the required objects to retrieve in this case adverts. From  HYPERLINK ""  , Google.com, facebook.com and youtube.com are ranked the best advertising websites globally. All the three are profit focused companies observing that youtube is still under Google, there for most of their advertising are not to the interest of the advertiser or user but to grow their businesses financially.
PROBLEM STATEMENT
Currently most businesses are embracing Internet advertising to market their products to masses of people using the Web, for instance in 2011 U.S spend more than $31.3 Billion [ HYPERLINK "/" ]; while a large numbers of internet users are searching for different market products and services. Statistics show that more than two billion people are using the Internet across the globe for different reasons, says a research by internet world statistics [ HYPERLINK "/stats.htm" /stats.htm]. Internet advertising revenue in the United States totaled $14.9 billion for the first six months of 2011: Search revenues accounted for 49%, display-related advertising accounted for 37%, classifieds revenues totaling to 8%, Lead Generation revenues accounted for 5%, email revenues accounted for 1% (Internet Advertising Revenue Report 2011). Advertisers want the best locations for their adverts as advert searchers hustle to get what they want. The problem escalates when an advertiser post an advert to a non-direct target market and market focus hence wasting resources, on the other hand, a lot of users suffer frustration as well time wastage since they might not be well versed with how search engines work to optimize their search process. One reason why TV adverts are still preferred to Internet advertisements is the fact that they are more glossier, self contained and do not compete for simultaneous editorial content. The proposed system aims to solve this dilemma by creating an advert pool that reduces the large volumes of information browsed when looking for an item and ranking the items.
On the other hand it allows users to suggest their advertising sites to be crawled for adverts hence ensuring their adverts are accessed by different product searchers.
GOALS AND Objectives:
Goal
To implement a system that simplifies advertisement search process of target market for sellers and ensure usefulness of Internet adverts to customers if easily accessed.
Objectives
To design and develop an advert crawler that searches for adverts across different on-line sites.
Rank adverts in order of their importance and relevance; e.g. location on the web page and number of similar words it has compared to search term.
Allow users to recommend best advertisement sites and sites with their adverts to ensure they are accessed.
Project Scope
The project is limited to demonstrate:
Reliability: Well structured data is necessary to ensure direct and aggregate results. Hence advert identification is the main area of focus to avoid direct but noise results to the user which is more like page information.
Completeness: Given data should be as complete as possible to achieve its objectives and offer trustworthy answers to the users.
Ranking Accuracy: Provide an optimal ranking mechanism critical for locating relevant object information since the information is in masses.
Scalability: Given the information to be extracted will grow enormously the system should achieve minimum scalability requirements using data processing technologies E.g. faster querying algorithms.
Relevancy: Since information keeps changing and web pages are continuously being updated, the system should frequently check on updates are re-index the sites.

JUSTIFICATION
The amount of information on the internet has increased manifold over the years. Users suffer from information overload, and most often the ever increasing information on web, increases demand on user’s time to browse through. The system is applicable to different products and services consumers, business individuals and companies to better market target research. The system increases vendor's sales, enhances choice making and reduces time used by product consumers to get relevant service and product specified at an appropriate price. When one is searching for an item, it is similar to window shopping where only the price and a summary of features are necessary. When you are selling for instance sports shoes you want people interested in sports or sports business to see your advert. The system should give an advertiser a chance to judge the relevancy of their advert on the recommended site, hence gain confidence on the location. The system reduces the amount of information browsed through on page compared to objects retrieved. Over time browsers have been improved to block some types of adverts like some flash banners which are assumed to be irritating ...