“字幕宝”

语言学家在掌握词汇方法VLS(Vocabulary Learning Strategies)上常年关注和推荐的是通过自然语言情境掌握(natural context)

字幕的四个境界

• 全中文
• 中外对照
• 全外语
• 只标注重点

• http://example.com/Batman.srt?level=cet4
• http://example.com/Batman.srt?level=cet6
• http://example.com/Batman.srt?level=gre

原字幕

720
00:46:44,369 --> 00:46:46,894
In that case, you could just step aside.

721
00:46:47,305 --> 00:46:48,465
Are you kidding me?

722
00:46:48,540 --> 00:46:51,031
You want us to step aside
after all we've done on this?

723
00:46:51,209 --> 00:46:53,939
We'd honor that.
We'd pay you a fee for your services.

724
00:46:54,012 --> 00:46:56,537
Say, $125 million?

725
00:46:57,215 --> 00:47:00,013
Plus an option to buy
a 10 percent stake in the company.

726
00:47:01,086 --> 00:47:02,417
Are you serious?

727
00:47:02,654 --> 00:47:06,090
I don't believe this!
You're offering me a fucking bribe?

728
00:47:07,158 --> 00:47:10,423
That's an offensive word,
offensive and untrue.

过滤后

720
00:46:44,369 --> 00:46:46,894
step aside: 靠边站

723
00:46:51,209 --> 00:46:53,939
fee: 费用

725
00:46:57,215 --> 00:47:00,013
stake: 股份

727
00:47:02,654 --> 00:47:06,090
bribe: 贿赂

728
00:47:07,158 --> 00:47:10,423
offensive(无理) and untrue(错误的).

DEMO

DEMO Code

• https://github.com/scateu/zimubao

进一步: 用户账户

• http://example.com/Batman.srt?user=scateu

更进一步: 训练集

• http://example.com/wordtest?user=scateu
• 用户在上面训练自己的词汇量，以使得过滤更精准

再进一步: 上接 OpenSubtitles.org

• OpenSubtitles - Developer
• OSDb协议
• XMLRPC of OpenSubtitles
• Timed track formats

商业模式

• 低买高卖
• 插入广告
• 与莆田系医院进行深度合作
• 不给钱就插入错的单词进行记忆污染勒索

商业模式示意

Participant Advertisement as AD
Participant User as U
Participant 字幕宝® as us
Participant Venture Capital as VC
U->us: Money
Note over us: Money burning
AD->us: A lot of money
us->U: Shit feeding
Note over U: Shit eating
Note over us: Money burning
VC->us: A lot of money
Note over us: So happy.

融资计划

• 天使轮: 5000人民币占股10%
  • 一次性全部买断股权可优惠到8折
• Pre A轮: $ 3,000,000,000

谢谢。

• Reference