|
|
On this page
I introduce the experiment purpose, the experiment conditions, the experiment method, and the analysis method of the experiment boiled, attached and conducted.
Since 's effect is what is greatly dependent on a visitor's action and psychology, even if it reasons an effect only on a desk, is it seldom persuasive?
Then, it experiments.
's effect was measured.
<<experiment purpose>>
It aimed at conducting the experiment operation of a trial production system is checked [ experiment ] and an effect is further proved [ experiment ] on 's research.
It experiments in this experiment on the Internet unavoidably not on the reaction enquiry to the announcement of invention, a publication, or invention but on a technical property.
Also you, readers,
"
When it applies, can a top page be expressed to the visitor introduced by the search engine as the ratio of which about?"
"After a top page is displayed, is there any visitor who is continuing the access further by the ratio of which about?"
. which I regard as being considered -- in this experiment, in order to reply to such a question, it aimed at measuring the effect of invention quantitatively
<<experiment condition>>
Two, the
Japanese site
of a "Sonobe Laboratory" and
this English site
, are used.
Comparison contrast of the behavior of the visitor before and behind 's “K—p was carried out.
A top page can say this site to . individual and the minor enterprise whose 1 page and 39 pages of non-top pages were as the site of the minor scale which often exists then, respectively.
The period which applied 's enbodiment is November 18, 2002 - 24.
About each site of Japanese and English, the application period and the non-applying period were taken one week each for comparison contrast, the period name was given as follows, and it considered as 2 site x2 period =4 sample.
-
AP : Application period (Monday, November 18 - Sunday, November 24)
-
NP : Non-applying period (December Monday,2nds - December Sunday, 8th)
., i.e., :, by which the name of a Japanese site and an English site sample should add the sign of J or E to each period name conclusion
-
APJ : Sample of the Japanese site of AP
-
APE : Sample of an English site of AP
-
NPJ : Sample of the Japanese site of NP
-
NPE : Sample of an English site of NP
It carries out.
<<experiment method>>
Know how applied to AP,
's -- they are the following realization (implementation)
One line is embedded in the notice at all the non-top's web pages so that a browser may call a CGI program to the visitor who has searched the Sonobe-Laboratory site at the time of an access and a part of display may be borne.
if search engines, such as Google, come out [ the page which introduced the access page ] and a certain thing can be distinguished, the CGI program will execute addition display dispensation by the server so that a top page may also be displayed in a new window in back if possible -- using a JavaScript language at the time of an access (Fig. 17)
[Fig. 17]
The example which indicated the top page and the advices message by addition
In addition, only the advices message by the letter and the picture was developed so that an access could be promoted also to the visitor who makes JavaScript the cancellation.
It is not influential especially when the page which introduced the access page is in a Sonobe-Laboratory site.
* Refer to the degree about a setup of effective/cancellation of JavaScript. :
* Clear explanation: How to enable/dispable JavaScript
(Sonobe Laboratory)
<<analysis method>>
The analysis methods of an experiment are all the web pages that constitute a Sonobe-Laboratory site, record the log of the information in connection with access, and carry out counting later.
Fig. 29 is a part of log. :
[Fig. 29]
A part of log
In this book, the part was applied black for protection of personal information.
About the log, in order to reduce an experimental error, the following amendments were performed. :
-
Access from the site of "crawler (crawler)" (namely, periodical enquiry program which a search engine site has) was removed.
-
Access to this external site performed when a visitor calls the translation function to foreign languages other than English from a translation site was removed.
-
Access by the experimenter itself was not recorded.
|
They are the notes which analyze a record flag and are issued in the log.
S is "a referring to agency is a search engine."
N is "a non-top page."
T is a "top page."
L is "clicking a link and moving outside."
C is "crawler."
H is "a referring to agency is a Sonobe-Laboratory home."
|
It is ‚ðŽ¦(ing).
When 's mechanism works, in a log, the record of T appears after [ of the record of the record flag SN ] several seconds.
An IP address or a user host is the address of a client, and if it approaches and the address looks at the same record, a visitor's actions of a series of are known.
Since Mr. cwod--, Mr. YahooBB22--, Mr. --musashi-tech--, and Mr. p2 --jp are after SN and the pattern T, for example, in this figure, respectively
's mechanism is operating.
-- It turns out that Mr. musashi-tech-- continues the access further.
Table 1 is material indicator statistics.
[Table 1]
<<material indicator statistics>>
Table 2 is access pattern statistics.
[Table 2]
<<access pattern statistics>>
In Table 2, the following symbol string shows each visitor's access pattern.
-
Signs T, t, and Nx are arranged in the right from the left in order of the time when access was executed.
-
Either of the signs s and r is attached to a head (example: sNtT).
-
Sign x or "-" may be attached to a conclusion. . (example: sN-)
-
T: The top page was accessed (T is the significance of top).
-
t: A server a top page
It displayed by 's •t‰Á display (t is the significance of top).
-
N: The non-top page was accessed. (N is the significance of non top) .
-
s: In order to show that a search engine is a referring to agency, attach to a head.
There is -, sT, and sN-. . (s is the significance of search)
-
d: In order to show that it is not Above s, there are .- and dT which are attached to a head, and dN- (d is the significance of direct).
-
x: This visitor's access has no it or later.
-
The symbol string which has not finished it as x contains all the patterns containing non-‚µ of access after it (example: in sN, sNT contains sNTT, sNTN, and sNTx further, including sNT, sNN, and sNx).
-
The symbol string which has finished with the sign "-" contains all the patterns containing non-‚µ after access which the pattern before "-" shows (example:-is the same as sN and sN).
|
If it says by this notation
A top page tends to be automatically taken out to the visitor "sN" who has searched ‚Í and the non-top page, and it is going to make it him at sNt.
Since operation of a JavaScript language was not however able to display a top page by his having been the visitor stopped by setup etc., it is not made to sNt, but although a ratio also has few cases set to sNx, sNN, etc., it exists.
It is verifiable objective with such enrollment and quantification.
The above
It is the scheme of an operative actual proof experiment.
I explain an analysis result to below.
|
|
