Honeydew0209

242 views

Published on

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
242
On SlideShare
0
From Embeds
0
Number of Embeds
15
Actions
Shares
0
Downloads
2
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Honeydew0209

  1. 1. An Update on the Honeydew Project Honeydew team
  2. 2. Basic Facts about Honeydew <ul><li>Existing honeydew system has </li></ul><ul><ul><li>Six components: hd, base, forms, plugins, viocore, viodata </li></ul></ul><ul><ul><li>7842 files and 2687 folders on disk </li></ul></ul><ul><ul><li>468 MB on disk </li></ul></ul>
  3. 3. Existing Temporal Expression Extractor <ul><li>TEA: </li></ul><ul><ul><li>Stands for Temporal Expression Anchorer </li></ul></ul><ul><ul><li>Ph.D. thesis of Benjamin Han at LTI </li></ul></ul><ul><ul><li>Three running modes </li></ul></ul><ul><ul><ul><li>'sentence‘: sentential mode </li></ul></ul></ul><ul><ul><ul><li>'tempex‘: temporal expression mode </li></ul></ul></ul><ul><ul><ul><li>'tcnl‘: Time Calculus for Natural Language mode. </li></ul></ul></ul><ul><ul><li>459 files, 23 folders, 71.2 MB on disk </li></ul></ul>
  4. 4. Baseline <ul><li>Modify TEA as our baseline to compare with Honeydew </li></ul><ul><li>Change TEA from the interactive running mode to the batch mode to process a large number of meeting emails without user interruption </li></ul>
  5. 5. Implementation <ul><li>Add Wrapper.py (done) </li></ul><ul><li>Change TimeShell.py (done) </li></ul><ul><li>Evaluation module (to be done by Wed) </li></ul>
  6. 6. Case Study <ul><li>Meeting email: </li></ul><ul><li>“ Yeah, this afternoon is probably better than tomorrow. </li></ul><ul><li>Let's say sometime around or after 2:00 p.m., ok? I need some time tothink about what I did. What do you say? </li></ul><ul><li>Guang” </li></ul>
  7. 7. Result from TEA <ul><li>---Sentence 0:<S>Yeah , <TEMPEX tcnl=&quot;{now + |0_{afternoon}|}&quot; time=&quot;( 20090209T13????..20090209T17???? )&quot; vcid=&quot;0&quot;>this afternoon</TEMPEX> <VC id=&quot;0&quot; ta=&quot;pres/none&quot;>is probably</VC> better than <TEMPEX tcnl=&quot;{now + |1_day|}&quot; time=&quot; 20090210 &quot; vcid=&quot;-1&quot;>tomorrow</TEMPEX> .</S>* VerbChunk: &quot;is probably&quot; (tense/aspect = pres/none)Time: &quot;this afternoon&quot; = {now + |0_{afternoon}|} = ( 20090209T13????..20090209T17???? )* VerbChunk: N/ATime: &quot;tomorrow&quot; = {now + |1_day|} = 20090210 </li></ul><ul><li>---Sentence 1:<S><VC id=&quot;0&quot; ta=&quot;pres/none&quot;>Let</VC> &apos;s <VC id=&quot;1&quot; ta=&quot;pres/none&quot;>say sometime</VC> around or <TEMPEX tcnl=&quot;{&gt; ^{14_hour, 0_min}}&quot; time=&quot;( 20090210T1401??..max )&quot; vcid=&quot;-1&quot;>after 2:00 p.m.</TEMPEX> , ok ?</S>* VerbChunk: &quot;Let&quot; (tense/aspect = pres/none)* VerbChunk: &quot;say sometime&quot; (tense/aspect = pres/none)* VerbChunk: N/ATime: &quot;after 2:00 p.m.&quot; = {> ^{14_hour, 0_min}} = ( 20090210T1401??..max ) </li></ul>
  8. 8. Data Collection <ul><li>Searched through my own meeting emails in my CS account </li></ul><ul><li>Selected 62 meeting emails out of 178 </li></ul>

×