holylovenia commited on
Commit
a589db4
1 Parent(s): 22ed505

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +53 -1
README.md CHANGED
@@ -5,4 +5,56 @@ tags:
5
  language:
6
  - jav
7
  - ind
8
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
  language:
6
  - jav
7
  - ind
8
+ ---
9
+
10
+ Sentiment analysis and machine translation data for Javanese and Indonesian.
11
+
12
+
13
+ ## Dataset Usage
14
+
15
+ Run `pip install nusacrowd` before loading the dataset through HuggingFace's `load_dataset`.
16
+
17
+ ## Citation
18
+
19
+ @article{Tho_2021,
20
+ doi = {10.1088/1742-6596/1869/1/012084},
21
+ url = {https://doi.org/10.1088/1742-6596/1869/1/012084},
22
+ year = 2021,
23
+ month = {apr},
24
+ publisher = {{IOP} Publishing},
25
+ volume = {1869},
26
+ number = {1},
27
+ pages = {012084},
28
+ author = {C Tho and Y Heryadi and L Lukas and A Wibowo},
29
+ title = {Code-mixed sentiment analysis of Indonesian language and Javanese language using Lexicon based approach},
30
+ journal = {Journal of Physics: Conference Series},
31
+ abstract = {Nowadays mixing one language with another language either in
32
+ spoken or written communication has become a common practice for bilingual
33
+ speakers in daily conversation as well as in social media. Lexicon based
34
+ approach is one of the approaches in extracting the sentiment analysis. This
35
+ study is aimed to compare two lexicon models which are SentiNetWord and VADER
36
+ in extracting the polarity of the code-mixed sentences in Indonesian language
37
+ and Javanese language. 3,963 tweets were gathered from two accounts that
38
+ provide code-mixed tweets. Pre-processing such as removing duplicates,
39
+ translating to English, filter special characters, transform lower case and
40
+ filter stop words were conducted on the tweets. Positive and negative word
41
+ score from lexicon model was then calculated using simple mathematic formula
42
+ in order to classify the polarity. By comparing with the manual labelling,
43
+ the result showed that SentiNetWord perform better than VADER in negative
44
+ sentiments. However, both of the lexicon model did not perform well in
45
+ neutral and positive sentiments. On overall performance, VADER showed better
46
+ performance than SentiNetWord. This study showed that the reason for the
47
+ misclassified was that most of Indonesian language and Javanese language
48
+ consist of words that were considered as positive in both Lexicon model.}
49
+ }
50
+
51
+
52
+ ## License
53
+
54
+ cc_by_3.0
55
+
56
+ ## Homepage
57
+
58
+ ### NusaCatalogue
59
+
60
+ For easy indexing and metadata: [https://indonlp.github.io/nusa-catalogue](https://indonlp.github.io/nusa-catalogue)