T. Kelly, "Thin-Client Web Access Patterns: Measurements from a Cache-Busting Proxy," Computer Comm. vol. 25, pp. 357-366, 2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....the log. In particular, the upper bound on entropy is given by H 0 (N) log 2 (N) and is attained when each object is equally likely to be referenced. It has been shown that the number of distinct references in a segment of a log increases with the size of the segment, even for very large logs [25]. Thus to be able to compare logs with different number of distinct references present, it is important to normalize the entropy measure. The appropriate normalization is based on the largest possible value of H(X) namely H 0 (N) Therefore the metric for popularity we will use is the normalized ....

T. Kelly, "Thin-client web access patterns: Measurements from a cachebusting proxy," in Proceedings of the 6th Web Caching Workshop, Boston, MA, 2002.

....the user does clicking on links, typing URLs, using navigational aids such as Back and Forward buttons and Bookmarks. It is also typically necessary to log at this level to capture activity that is served by the browser cache, although one potential alternative is to use a cache busting proxy [Kel01] Using an individual client history to build a model of the client provides the opportunity to make predictions that are highly personalized, and thus reflect the behavior patterns of the individual user. Unfortunately, logs from augmented browsers (such as those captured by Tauscher and ....

Terence Kelly. Thin-client Web access patterns: Measurements from a cache-busting proxy. In Web Caching and Content Delivery: Proceedings of the Sixth International Web Content Caching and Distribution Workshop (WCW'01), Boston, MA, June 2001.

....cache hit. Moreover, a trace collected at a proxy normally fails to capture any user requests that hit in the client (browser) caches. Other problems with conventional proxy logs include inadequate detail, low resolution timestamps, and poor clock synchronization in multiple host proxy arrays [12, 15, 17, 28]. In principle, one could avoid such problems by collecting traces using an instrumented client; this could capture every user reference. Instrumented browsers have been used to collect traces from small user populations [14, 16] It is difficult to instrument popular browsers today because ....

.... Since the source code for the proxy was available, WebTV could instrument it to collect data not ordinarily logged (e.g. payload digests, which have on rare occasions been logged by modified proxies in the past [35, 36] For more discussion of WebTV s trace collection methodology, see Reference [28]. A cache busting proxy allowed WebTV to collect a very large trace from a large user population at reasonable cost. The WebTV trace is comparable in size to the largest traces used in HTTP namespace investigations. Unlike these proxy, sniffer, server, and crawler traces, however, the WebTV ....

[Article contains additional citation context not shown here]

T. Kelly. Thin-client Web access patterns: Measurements from a cache-busting proxy. Computer Communications, 25(4):357--366, Mar. 2002. http: //ai.eecs.umich.edu/tpkelly/papers/.

.... at Georgia Tech, and used it to study methods of user interaction with the browser as well as user navigation patterns [8] Kelly has recently collected a very large anonymized client trace from WebTV, and has used it to explore browser cache performance, reference locality, and document aliasing [23]. All of these studies have used passive measurement to explore user behavior and client performance, whereas we use the Medusa proxy to compare the performance of various Web delivery systems using active measurement. Krishnamurthy and Wills [26] studied end to end Web performance from a ....

T. Kelly. Thin-Client Web Access Patterns: Measurements from a Cache-Busting Proxy. In Proc. of the 6th Int. Web Caching and Content Delivery Workshop, June 2001.

....cache hit. Moreover, a trace collected at a proxy normally fails to capture any user requests that hit in the client (browser) caches. Other problems with conventional proxy logs include inadequate detail, low resolution timestamps, and poor clock synchronization in multiple host proxy arrays [15, 12, 17, 28]. In principle, one could avoid such problems by collecting traces using an instrumented client; this could capture every user reference. Instrumented browsers have been used to collect traces from small user populations [14, 16] It is difficult to instrument popular browsers today because ....

.... Since the source code for the proxy was available, WebTV could instrument it to collect data not ordinarily logged (e.g. payload digests, which have on rare occasions been logged by modified proxies in the past [35, 36] For more discussion of WebTV s trace collection methodology, see Reference [28]. A cache busting proxy allowed WebTV to collect a very large trace from a large user population at reasonable cost. The WebTV trace is comparable in size to the largest traces used in HTTP namespace investigations. Unlike these proxy, sniffer, server, and crawler traces, however, the WebTV ....

[Article contains additional citation context not shown here]

T. Kelly. Thin-client Web access patterns: Measurements from a cache-busting proxy,. Computer Communications, 25(4):357--366, Mar. 2002. http: //ai.eecs.umich.edu/tpkelly/papers/.

No context found.

T. Kelly, "Thin-Client Web Access Patterns: Measurements from a Cache-Busting Proxy," Computer Comm. vol. 25, pp. 357-366, 2002.

No context found.

T. Kelly. Thin-client Web access patterns: measurements from a cache-busting proxy. In Proc. of the Int. Workshop on Web Content Caching and Distribution, 2001.

No context found.

T. Kelly. Thin-client Web access patterns: measurements from a cache-busting proxy. In Proc. of the Int. Workshop on Web Content Caching and Distribution, 2001.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC