ANNOTATING WEB SEARCH RESULTS

Abstract

With more than millions of pages, the Web has become a greatly enormous information source. This information is in form of documents, images, videos as well as text. With such vast sizes of data, it is a common problem to get the right information that one wants. Oftentimes users have to search for the right content they are looking for from the Web with the help of search engines. Searching can be done manually by use of available platforms like Google or automatically in form of web crawlers. Since the semantic web is not structured, search results can include varying types of information relating to the same query. Sometimes these results cannot be directly analyzed to meet the specific interpretation need. The search result records (SRRs) returned from the Web following manual or automatic queries are in form of web pages that hold results obtained from underlying databases. Such results can further be used in many applications such as data collection, comparison of prices etc. Thus, there is a need to make the SRRs machine processable. To achieve that, it is important that the SRRs are annotated in a meaningful fashion. Annotation adds value to the SRRs in that the collected data can be stored for further analysis and makes the collection easier to read and understand. Also annotation prepares the data for data visualization. The SRRs bearing same concepts are grouped together thus making it easier to make comparisons and analyze and go through the collection. The purpose of this research is to find out how search results from the Web can be automatically annotated and restructured to allow for data visualization for users in a specific domain of discourse. A case study application is implemented that uses a web crawler to retrieve web pages about any topic in public health domain. This research is a continuation of the work done by Mr. Emanuel Onu in the project “Proposal of a Tool to Enhance Competitive Intelligence on the Web”.

1 Introduction

People of all walks of life use the internet for so many different tasks such as buying and selling items, social networking, digital libraries, news, etc. Researchers need information from digital libraries and other online document repositories to conduct their research and share information; scholars need books to get information and knowledge from; people communicate to one another through emails via the Web; others utilize social media to exchange information as well as having casual chat; some conduct transactions like purchasing items and paying for bills via the web. The World Wide Web is today the main “all kind of information” repository and has been so far very successful in disseminating information to humans. The Web has become the preferred medium for many database applications, such as e-commerce and digital libraries. Many database applications store information in huge databases that users’ access, query and update through the Web. The improvement in hardware technologies has seen increase in computers and server’s storage capacity. As such, many web servers store a lot of data in their storage drives. In some social media websites e.g. Facebook[1], users can upload pictures, videos as well as other documents. YouTube [2] allows its users to post videos of varying lengths to their servers. There are other automated systems that collect a lot of data on daily basis. For example, bank systems need to store daily Auto Teller Machine (ATM) transactions as well as other customers’ transactions. Some monitoring systems collect data about some aspect of life e.g. climate change, online shopping systems that keep information about the clients’ daily shopping experience. These are some but few ways that have led to a gigantic amount of information and documents to be available on the Web.

However, due to the heterogeneity and the lack of structure of Web information sources, access to this huge collection of information has been limited to browsing and searching. That is, for one to access a document, you need to put the URL (Universal Resource Locator) in the Web Browser or making use of a search engine. The former way is suitable when you know what you are looking for and the exact location on the Web. But this is hardly the case and as such many of the Web users locate particular content they are looking for by using search engines. There are software systems that require a user to manually enter a search term and the search engine retrieves documents according to the term entered by the user; while there are also other automated search engines that make use of a Web Crawler.

There are several notable web based search engines that index web documents and are available for use by Web users. Most common ones are Google, Yahoo, AltaVista and many more. Such systems search through the collection of the documents sourced from the Surface Web – which is indexed by standard search engines as well as the Deep Web –which requires some special tools to be accessed. Most users benefit from such systems when researching information that is not known or they want to redirect to trace a website they know but can’t remember its URL. Still, there are some business disciplines such as Competitive Intelligence [3] that require particular type of information (domain specific) in order to make strategic business decisions. In such scenarios, different tools are developed in order to help in information gathering and analysis. Several other methods for searching and information retrieval to gather intelligence also work in such domain specific areas. For example, manually browsing the Internet could be the simplest method for conducting a Competitive Intelligence task. Manual browsing of the Internet to a reasonable level guarantees the quality of documents collected which in turn improves the quality of knowledge that is discoverable.[4] However, the challenge here is that a lot of time is spent. According to Onu, a survey of over 300 Competitive Intelligence professionals shows that data collection is the most time-consuming task in typical Competitive Intelligence projects, amounting to more than 30% of the total time spent in the whole project. In this case, for Competitive Intelligence professionals to manually browse the Internet to read the information on every page of a Website in order to locate useful information and also to synthesize the information, it is mentally exhausting and overwhelming. There is undeniably a huge demand for collecting data of interest from multiple Websites across the Web. For example, an online shopping system that collects multiple result records from different item sites, there is a need to determine whether any two items retrieved in the search result records refer to the same item. For a book online shopping system, the ISBN can be compared to achieve this. If ISBNs are not available, then their titles and authors could be used instead. Such a system is also expected to list down the price of an item from each site. Thus the system needs to know the semantic of each data unit. Unfortunately, the semantic labels of data units are often not provide in the result pages. For instance, in Figure X, no semantic labels for the values of title, author, publisher, etc., are given. Having semantic labels for data units is not only important for the above record linkage task, but also for storing collected search result records into a database table (e.g., Deep web crawlers) for later analysis. Early applications require tremendous human efforts to annotate data units manually, which severely limit their scalability.

Different tools have been developed that help to search, gather, analyze, categorize, and visualize a large collection of web documents. One of such tools is one proposed by Mr. Emmanuel Onu in his paper “Proposal of a Tool to enhance Competitive Intelligence on the Web”, and the tool is called the CI Web Snooper. It is from this tool that this paper is based. This research is a continuation of the initial work from the above mentioned paper. The CI Web Snooper is a tool for searching and retrieving Websites from the Internet that can be used for information gathering and knowledge extraction. It uses a real-time search technique so that the information it sources from the Web is update. It has four major components: User Interface, Thesaurus Model, Web Crawler and Indexer. The User Interface allows the user to specify search query and also specify seed URLs for the Web Crawler to use in its search. The Thesaurus Model is used to model the domain of interest and is key to query reformulation and indexing of Web pages. The Web Crawler component is responsible for finding and downloading Web pages using Breadth-First search algorithm that starts from the URLs specified by the user.

Department: Computer Science (M.Sc Thesis)

Format: MS Word

Chapters: 1 - 5, Preliminary Pages, Abstract, References, Appendix.

No. of Pages: 34

Price: 20,000 NGN

In Stock

Buy Now

Our Customers are Happy!!!

See What Some of Our Customers are Saying About Us