-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathREADME
68 lines (50 loc) · 2.68 KB
/
README
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
Version 1.0
GoldStandard Word-Polarity
-----------------------------------------------------------------------------
This GoldStandard_Word_Polarity list has been created by Shachi H Kumar, from
the GoldStandard annotated data developed by Caecilia Zirn et al. and used in
the paper :
"Fine-Grained Sentiment Analysis with Structural Features".
Please cite the above paper if this dataset is used.
Dataset :
The data is obtained from the Multi-Domain Sentiment Dataset that contains
product reviews of many product types taken from amazon.com.
Three product categories have been used as a part of this GoldStandard data
generation - Kitchen and Houseware, Gourmet Food, Cellphones and Services.
The reviews have been marked as positive or negative based on the star ratings.
The raw_reviews folder consists of 20 longest positive and negative reviews
for each of the product categories.
The reviews have been numbered as per their usage in the GoldStandard.
Each review was annotated by 3 independent annotators. They were asked to mark arbitrary parts of the reviews as positive, negative, positive_towards_other_product, negative_towards_other_product or other, with the minimal units being words.
Then, the majority label for each word is derived.
The GoldStandard_WordList folder consists of the following fields :
ReviewID, Word, and Polarity.
The ReviewID field indicates the review each word belongs to. The reviewID
also corresponds to the reviewIDs of the raw_reviews.
The Word column gives each word from the review, and the Polarity column
gives the goldStandard polarity of each word, taken as the majority polarity
of the annotations done by the annotators.
Please find more details of this in
the "Fine-Grained Sentiment Analysis with Structural Features" paper.
It has five main sentiments.
1) Positive
2) Negative
3) Other
4) negative_other_product (Mentions of negative aspects of other products)
5) positive_other_product (Mentions of positive aspects of other products)
.......................................................................
PUBLICATIONS
------------
As mentioned above, the data and its annotation is described in more detail
in the the paper.
Links to the paper is available here:
http://anthology.aclweb.org//I/I11/I11-1038.pdf
.......................................................................
For more information on the data , contact :
-------------------------------------------------------------------------
Caecilia Zirn
Shachi H Kumar
Graduate Student at UCSC
email: [email protected]
.......................................................................