Statistics
| Branch: | Tag: | Revision:

root / NEWS @ 2d9cf61c

History | View | Annotate | Download (1.74 KB)

1
Overview of Changes
2
===================
3

    
4
* 0.7 (2012/01/13)
5

    
6
    * External LibSylph (standalone or Sylpheed) can be used instead of
7
      built-in one.
8

    
9
* 0.6 (2011/11/14)
10

    
11
    * Received header is also used for classification now.
12
    * The verbose mode has several levels now.
13
    * Sylfilter command: an option to specify database directory was added.
14
    * Sylfilter command: -V (print version) option was added.
15

    
16
* 0.5 (2011/10/13)
17

    
18
    * The calculation of combined probability of Robinson-Fisher method was
19
      fixed. The version 0.4 wrongly applied the total number of words in
20
      a message without excluding the words within the minimum deviation.
21
      This fix will introduce more accurate filtering and a small speedup.
22
    * Sylfilter command: options to specify Robinson-Fisher parameters were
23
      added.
24
    * The bug that a multibyte filename could not be passed on Windows was
25
      fixed.
26

    
27
* 0.4 (2011/10/6)
28

    
29
    * Robinson-Fisher method was implemented, and it is used by default.
30
      This will improve the accuracy of filtering. Previous Paul/Naive bayes
31
      method still can be used with -m option.
32
    * Sylfilter command: an option to specify filtering method (-m) was added.
33

    
34
* 0.3 (2011/9/29)
35

    
36
    * Sylfilter command: -j and -C / -c and -J option can be specified
37
      at the same time
38
    * Sylfilter command: -B (no-bias for clean messages) option was added.
39
    * GDBM support was added (exclusive with QDBM)
40
    * APIs for global configuration were added
41

    
42
* 0.2 (2011/9/13)
43

    
44
    * Speedup learning (data incompatible to 0.1)
45
    * Support Windows
46

    
47
* 0.1 (2011/9/5)
48

    
49
    * Initial release
50
    * Supports QDBM and SQLite
51
    * Provides libsylfiter library and sylfilter command-line tool
52
    * Initial implementation of bayesian filter