Understanding User Intentions by
Computational Techniques
Hongning Wang
Department of Computer Science University of Illinois at
Urbana-Champaign Urbana IL, 61801 USA
Research Summary
Great hotel = price<$60,
location in downtown
Good MP3 player = large
memory, long batter life
Relevant news = most
recent report
 Latent Aspect Rating Analysis [KDD’10, 11]
 Online Forum Discussion Structure Modeling [SIGIR’11]
 Latent Topical Structure Modeling [ACL’11]
10/7/2015
2
Information buried in text content
10/7/2015
3
Latent Aspect Rating Analysis [KDD’10, 11]
Entity
Aspects
Review
Aspect Rating Aspect Weight
Location
location
amazing
walk
anywhere
Room
room
dirty
appointed
smelly
Service
terrible
front-desk
smile
unhelpful
10/7/2015
Excellent location in walking
distance to Tiananmen Square and
shopping streets. That’s the best
part of this hotel! The rooms are
getting really old. Bathroom was
nasty. The fixtures were falling off,
lots of cracks and everything
looked dirty. I don’t think it worth
the price. Service was the most
disappointing part, especially the
door men. this is not how you treat
guests, this is not hospitality.
0.86
0.04
0.10
4
LARA Applications I
• User rating emphasis analysis
City
AvgPrice
Amsterdam
241.6
San Francisco
261.3
Florence
272.1
Group
Val/Loc
Val/Rm
Val/Ser
top-10
190.7
214.9
221.1
bot-10
270.8
333.9
236.2
top-10
214.5
249.0
225.3
bot-10
321.1
311.1
311.4
top-10
269.4
248.9
220.3
bot-10
298.9
293.4
292.6
– Reviewers emphasize ‘value’ aspect would prefer
‘cheap’ hotels
10/7/2015
5
LARA Applications II
• User rating behavior analysis
Expensive Hotel
Cheap Hotel
5 Stars
3 Stars
5 Stars
1 Star
Value
0.134
0.148
0.171
0.093
Room
0.098
0.162
0.126
0.121
Location
0.171
0.074
0.161
0.082
Cleanliness
0.081
0.163
0.116
0.294
Service
0.251
0.101
0.101
0.049
– Reviewers focus differently on ‘expensive’ and ‘cheap’
hotels
6
Oops, time is limited...
Online Forum Discussion
Structure Modeling [SIGIR’11]
Latent Topical Structure
Modeling [ACL’11]
• Probabilistic model with
rich features
• Topical transition structure
Initial
topic
Content topic
proportion
10/7/2015
Transition
probability
Emission
probability
7
Future Direction
• Mining rich user-generated-data
– Clicks, sharing, like
• Analyzing social interactions
– Friendship, following
10/7/2015
8
Thank you!
• Q&A
10/7/2015
9
Information Hidden in Structures
• Structure is not always visible
Flat View
Threaded View
v.s.
10
10/7/2015
Online Forum Discussion Structure
Modeling
10/7/2015
11
Online Forum Discussion Structure Modeling
• Probabilistic model with rich features
– Post attributes
– User interactions
10/7/2015
12
Recognizing and modeling document structure
• Languages are intrinsically cohesive and
coherent
10/7/2015
13
Latent Topical Structure Modeling
Initial topic
Content topic
proportion
10/7/2015
Transition
probability
Emission
probability
14
LARA Applications I
• Corpus specific word sentimental orientation
Value
Rooms
Location
Cleanliness
resort 22.80
view 28.05
restaurant 24.47
clean 55.35
value 19.64
comfortable 23.15
walk 18.89
smell 14.38
excellent 19.54
modern 15.82
bus 14.32
linen 14.25
worth 19.20
quiet 15.37
beach 14.11
maintain 13.51
bad -24.09
carpet -9.88
wall -11.70
smelly -0.53
money -11.02
smell -8.83
bad -5.40
urine -0.43
terrible -10.01
dirty -7.85
road -2.90
filthy -0.42
overprice -9.06
stain -5.85
website -1.67
dingy -0.38
▫ Uncover sentimental information directly from the
data
15
Learned Topical Structure
10/7/2015
16
Descargar

Understanding User Intentions by Computational Techniques