Exercise 14: Searching mechanisms, Virtual worlds and Cyberagents

1. What is a spider? What does it do?

A spider is a web based program which harvests information for search engines and other allied sites (Ince, 2004).Spiders are typically programmed to visit sites that have been submitted by their owners as new or updated. Entire sites or specific pages can be selectively visited and indexed.Spiders visit Web sites, record the information there, read the meta tags that identify a site according to subjects, and follow the site's links to other pages. Because of the many links between pages, a spider can start at almost any point on the Web and keep moving. Eventually it returns the data gathered on its journey to the search engine's central depository of information, where it is organised and stored. Periodically the spider will revisit the sites to check for changed information, but until it does so, the material in the search engine's index remains the same. It is for this reason that a search at any time may yield "dead" Web pages, or ones that can no longer be found. Spiders are called spiders because they usually visit many sites in parallel at the same time, their "legs" spanning a large area of the "web.

2. Differentiate the various types of software agents.

There are a numnber of different types of software agents that have been developed. Ince (2004) identifies and defines a number of these:

• Chatterbots. A chatterbot is an agent which converses with the user in some restricted form of natural language. For example a bot of this type may retrieve a web site in response to a natural language question.
  

• Commerce Agents. These carry out large scale, e-commerce functions as distinct from the more specialised agents such as electronic shopping agents. A typical example of a commerce agent is one which mediates between two companies who have a business relationship, this would perform processes such as reconciling invoices, notifying product arrival etc.

• Data Management Agents. These are agents which carry out tasks associated with corpora of data. For example, processing the text on a web site and providing a summary of that text with keywords and keyphrases.

• Government Agents. These are agents which carry out some activity related to the government such as searching for government regulations or extracting government statistics. Most of these agents are American.

• News Agents. A news agent is concerned with tasks that are associated with Web Based news services. Some functions that these agents may perform are:

• Displaying breaking news on your desktop as it is notified to the web site which contains the news.
• Scan news sites for articles of interest to the user.
• Send an email when a news story on a particular interest is published.
• Develop a personalized newspaper which contains only stories of interest to the user.

• Newsgroup Agents. These are agents which carry out some tasks with Internet newsgroups. Some typical tasks are:
  

• Filtering posts to match keywords
• Scanning and prioritizing postings
• Notifying users of posted URL's
• Searching newsgroups for pictures and displaying them

• Shopping Agents.These are used to carry out tasks associated with the accessing of retail sites. Examples include:

• • Finding the cheapest prices for a product.
  • Scan bid prices and auction times.

• Software Agents. These are agents which carry out tasks associated with software eg. notifying when updates are available, downloading updates, installing components etc.

• Stock Agents. These agents are associated with the purchasing and selling of stocks and shares.
  

• Update Agents. Which notifies a user when a change has occurred which is of significance to the user. Exmaples of these are:
  
  
  • Alerting user when news has been posted on a website about a company.
  • Alerting a user when a book published by a particular becomes available.
  • Reporting when a particular web page has changed.
  • Announce when a particular job becomes available on a job board.

• Web Developer Agents. Which are used by webmasters to carry out their job. The most popular agent of this type is one to check links on a website to identify if they are active or dead. Other examples of functions which these agents may perform are:
  
  
  • Track statistics like page hits, searches etc.
  • Caching pages
  • Identifying if users are trying to illegally access a restricted page/site etc

3. Identify various activities in e-commerce where software agents are currently in use.

As stated in Fasli (2007) software agents are computational systems that are capable of autonomous, reactive and proactive behaviour, and are also able to interact with each other. Whilst agents and "bots" have been employed to undertake malicious behaviour, the application of software agents in e-Commerce is one of the fastest-growing and most exciting areas of computer science. This emerging technology is enabling individuals and businesses to take advantage of the new and powerful medium of the World Wide Web. Some of the various activities in e-commerce where software agents are currently in use are as follows:

• Using shopping agents to trawl sites for the cheapest price of a particular product.
• Emailing a user when products/jobs etc become available.
• Notifying users when new products are released.
• Create custom news/product/topic listings based on personal preferences.

4. Computing ethics and bot programming case study: rocky

a. Get an account username and password from the lecturer to LC_MOO at http://ispg.csu.edu.au:7680 and login to the Welcome Lobby.

Currently waiting for login and password from the lecturer in order to complete this part of the exercise.Watch this space.

Update: Have now recieved login from Lecturer and have spent five minutes chatting with Rocky. Update below.

b. Hold a 5-minute discussion with Rocky on a special topic.
Commands and chat are entered in the command box (bottom-left of
screen in Figure 11) : act rocky (start bot) hush rocky (stop bot)

c. Rocky is an ELIZA-like bot. Report your findings.
.

In the absence oif a login to LC_MOO, I decided to have a look around in the internet and find the original ELIZA bot (http://nlp-addiction.com/eliza/) and have a 5 minute conversation with Eliza ELIZA is a natural language conversation program described by Joseph Weizenbaum in January 1966. It features the dialog between a human user and a computer program representing a mock Rogerian psychotherapist.As such, her responses are designed to keep you talking and get to the root of your deep seated pyshcological problems (apparently). Obviously Eliza is a little dated now, as a lot of her responses where repetitive, but even so, it was interesting to experiment with different responses. Following playing around with this, I am keen to get in and have a discussion with Rocky, given he is a little more modern.

After gaining access to LC_MOO I typed "Act Rocky" to start Rocky, and proceeded to have a discussion with Rocky. I have to admit, conversation with Rocky wasn't overly stimulating. I have attached a screen capture below which shows part of my conversation with Rocky. As can be seen from this, whenever Rocky was confronted with a question he didn't undertand he would either outright state that he did not understand the question, or he would turn the question back on me and ask me something along the lines of "Why Do you want to know that". At one point he asked me if I could program in Python, but beyond that the discussion didn't go anywhere. I asked him his name, and he didn't understand the question. I asked him the date and he didn't understand the question. From my brief discussion with Rocky, it would appear that he is still at a very rudimentary level, able to idnetify basic questions and offer a basic response if he understands the question, or explaining that he doesn't understand the question if this is the case. After playing with ELIZA, as per the previous paragraph, I have to admit, I was expecting a higher level of complexity with Rocky.