How It Developed The story so far Why I Wrote It Reasons... How It Works Some insights

How It Developed 1.0beta build 0003 12 December 1999 Created whole installation kit; added forgotten definition files tld.txt and httpStatus.txt; added reports for number of sessions, total duration of sessions, average time per session, distribution of hits over the week 1.0beta build 0002 9 December 1999 Included error messages if definition files can't be found 1.0beta build 0001 8 December 1999 First public version! Why I Wrote It Recently I wondered what the visitors on my website are actually doing. What I wanted was some analysis of their behaviour when they're browsing through the pages. I looked at a couple of log file analysers and found they either didn't provide what I needed or their prices are pretty steep. So I pulled Perl out of the tool box and wrote AC.log which can be downloaded here for free - "free" because I want to give back to the online community. How It Works A line from a log file looks like this: From this information it's fairly easy do derive the standard reports such as most downloaded files or the distribution of downloads over the day. More interessting though is the analysis of browsing patterns. For this we have to look at sessions. On which page did a user session begin, which path did the user follow through the web site, how long did s/he look at a particular page, and on which page did the session end. Problem: Since HTTP is a protocol that establishes (and terminates) a connection for each retrieved item, be it a HTML page, a GIF icon, or a JPEG image, there is no such thing as a session - at least not in HTTP. This implies there is no direct way to tell which pages were downloaded in one session. Solution: AC.log assumes if no more than a certain time has passed between two accesses from the same host, these actions are related to each other and signify they have been made by one person in one session.