Fix partially hidden article

Fix partially hidden article

Score changes:

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/top-mobile-mhtml.html
  One entry changed.
    Precision: 0.970 → 0.994
    Recall:    0.204 → 0.991
  Average precision: 0.903 → 0.905
  Average recall:    0.888 → 0.949

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/knowledge.html
  One entry changed.
  Average precision: 0.961 → 0.960

BUG=599121
R=mdjones@chromium.org

Committed: d16a68c1b885f679ae1266a9eeee66e69c12e67c

[wychen, 2016-08-24 20:17:30]

wychen@chromium.org changed reviewers:
+ mdjones@chromium.org

[wychen, 2016-08-24 20:17:30]

PTAL. This is more like a specific hack.

[mdjones, 2016-08-24 22:45:51]

https://codereview.chromium.org/2267403008/diff/1/java/org/chromium/distiller...
File java/org/chromium/distiller/webdocument/DomConverter.java (right):

https://codereview.chromium.org/2267403008/diff/1/java/org/chromium/distiller...
java/org/chromium/distiller/webdocument/DomConverter.java:96: &&
DomUtil.hasClassName(e, "hidden")) {
Do you think it is possible/worth while to isolate this code into an extractor?
You could have it hold a reference to hiddenElements and keep track of it that
way.

[wychen, 2016-08-25 15:32:00]

Speed profiling on reader-mode-golden-data.sstable

master: Time = 102.527969 +- 0.071651 ms, averaged over 1000 runs
ps1:    Time = 102.553285 +- 0.070135 ms, averaged over 1000 runs
ps2:    Time = 102.515819 +- 0.070401 ms, averaged over 1000 runs

[mdjones, 2016-08-29 18:27:37]

On 2016/08/25 15:32:00, wychen wrote:
> Speed profiling on reader-mode-golden-data.sstable
> 
> master: Time = 102.527969 +- 0.071651 ms, averaged over 1000 runs
> ps1:    Time = 102.553285 +- 0.070135 ms, averaged over 1000 runs
> ps2:    Time = 102.515819 +- 0.070401 ms, averaged over 1000 runs

How does this affect our precision/recall?

[wychen, 2016-08-29 18:45:10]

On 2016/08/29 18:27:37, mdjones wrote:
> On 2016/08/25 15:32:00, wychen wrote:
> > Speed profiling on reader-mode-golden-data.sstable
> > 
> > master: Time = 102.527969 +- 0.071651 ms, averaged over 1000 runs
> > ps1:    Time = 102.553285 +- 0.070135 ms, averaged over 1000 runs
> > ps2:    Time = 102.515819 +- 0.070401 ms, averaged over 1000 runs
> 
> How does this affect our precision/recall?

In our existing datasets, one entry in golden_data_with_knowledge is affected.
That one happens to use class="hidden" to hide a div that should actually be
hidden.

For the top_mobile_mhtml dataset in construction, no changes except for our
target.

I'll add these results to the description when top_mobile_mhtml is done.

[wychen, 2016-08-30 01:36:23]

Description was changed from

==========
Fix partially hidden article

BUG=599121
==========

to

==========
Fix partially hidden article

Score changes:

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/top-mobile-mh...
  One entry changed.
    Precision: 0.970 → 0.994
    Recall:    0.204 → 0.991
  Average precision: 0.903 → 0.905
  Average recall:    0.891 → 0.952

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/knowledge.html
  One entry changed.
  Average precision: 0.961 → 0.960

BUG=599121
==========

[wychen, 2016-08-30 01:37:05]

On 2016/08/29 18:45:10, wychen wrote:
> I'll add these results to the description when top_mobile_mhtml is done.

Updated the scores in the description.

[wychen, 2016-08-30 01:56:59]

Description was changed from

==========
Fix partially hidden article

Score changes:

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/top-mobile-mh...
  One entry changed.
    Precision: 0.970 → 0.994
    Recall:    0.204 → 0.991
  Average precision: 0.903 → 0.905
  Average recall:    0.891 → 0.952

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/knowledge.html
  One entry changed.
  Average precision: 0.961 → 0.960

BUG=599121
==========

to

==========
Fix partially hidden article

Score changes:

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/top-mobile-mh...
  One entry changed.
    Precision: 0.970 → 0.994
    Recall:    0.204 → 0.991
  Average precision: 0.903 → 0.905
  Average recall:    0.888 → 0.949

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/knowledge.html
  One entry changed.
  Average precision: 0.961 → 0.960

BUG=599121
==========

[mdjones, 2016-08-30 15:15:39]

lgtm

[wychen, 2016-08-30 16:55:39]

Description was changed from

==========
Fix partially hidden article

Score changes:

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/top-mobile-mh...
  One entry changed.
    Precision: 0.970 → 0.994
    Recall:    0.204 → 0.991
  Average precision: 0.903 → 0.905
  Average recall:    0.888 → 0.949

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/knowledge.html
  One entry changed.
  Average precision: 0.961 → 0.960

BUG=599121
==========

to

==========
Fix partially hidden article

Score changes:

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/top-mobile-mh...
  One entry changed.
    Precision: 0.970 → 0.994
    Recall:    0.204 → 0.991
  Average precision: 0.903 → 0.905
  Average recall:    0.888 → 0.949

https://x20web.corp.google.com/~wychen/domdistillerscore/mobile/knowledge.html
  One entry changed.
  Average precision: 0.961 → 0.960

BUG=599121
R=mdjones@chromium.org

Committed: d16a68c1b885f679ae1266a9eeee66e69c12e67c
==========

[wychen, 2016-08-30 16:55:41]

Committed patchset #2 (id:20001) manually as
d16a68c1b885f679ae1266a9eeee66e69c12e67c (presubmit successful).