Academia.eduAcademia.edu

Figure 3 – uploaded by Swapan Debbarma

See full PDFdownloadDownload figure
Fig. 3 Flow chart of proposed method The polarity information of any given word of the dictionary used as a feature set. After calculating the system perfor- mances, computational analysis is done on the results. We performed linear extrapolation of the data taken by both the feature set and found that for the dictionary, and TS features seem to converge, at 52% and 39% respectively in the limits of the number of songs goes to infinity. It has been shown in Fig. 3. The objective of this paper is to start this work and expand slowly to build a more extensive database for build- ing the sentimental dictionary since we just discovered in this paper that at present dictionary gives better performance than TS feature.

Figure 3 Flow chart of proposed method The polarity information of any given word of the dictionary used as a feature set. After calculating the system perfor- mances, computational analysis is done on the results. We performed linear extrapolation of the data taken by both the feature set and found that for the dictionary, and TS features seem to converge, at 52% and 39% respectively in the limits of the number of songs goes to infinity. It has been shown in Fig. 3. The objective of this paper is to start this work and expand slowly to build a more extensive database for build- ing the sentimental dictionary since we just discovered in this paper that at present dictionary gives better performance than TS feature.

Related Figures (15)

Fig. 2. Havner’s mood taxonomy Fig. 1 Russell’s mood taxonomy
Fig. 2. Havner’s mood taxonomy Fig. 1 Russell’s mood taxonomy
Fig.4 Creation of sentimental word dictionary from the holy bible
Fig.4 Creation of sentimental word dictionary from the holy bible
Fig.5 Creation of polarity annotated dataset for classification
Fig.5 Creation of polarity annotated dataset for classification
Table 1 Text Stylistic Features
Table 1 Text Stylistic Features
Fig.6 Features taken from the sentimental word dictionary In this section of the paper, we developed a classification system for Kokborok song. In the system, lyrics are clas- sified by three types of polarities (positive or negative or neutral) that were assigned to a song after the interpretation of its consequent lyric. We create two classification systems separately, one is using only TS Features, and another is by mapping the dictionary words to a particular song. Process of features taken from dictionary is shown in Fig. 6. For
Fig.6 Features taken from the sentimental word dictionary In this section of the paper, we developed a classification system for Kokborok song. In the system, lyrics are clas- sified by three types of polarities (positive or negative or neutral) that were assigned to a song after the interpretation of its consequent lyric. We create two classification systems separately, one is using only TS Features, and another is by mapping the dictionary words to a particular song. Process of features taken from dictionary is shown in Fig. 6. For
Average accuracy rate 41% Table 2 Performance Evaluation for TS Feature
Average accuracy rate 41% Table 2 Performance Evaluation for TS Feature
Fig.8 Comparison of accuracy rate by different algorithm for dic- tionary features
Fig.8 Comparison of accuracy rate by different algorithm for dic- tionary features
Fig. 7 Comparison of accuracy rate by different algorithm for TS fea- ture
Fig. 7 Comparison of accuracy rate by different algorithm for TS fea- ture
Table 3. Performance Evaluation for Dictionary Features Average accuracy rate 56%
Table 3. Performance Evaluation for Dictionary Features Average accuracy rate 56%
Table 4 Confusion matrix for TS features
Table 4 Confusion matrix for TS features
Table 5 Confusion matrix for Dictionary feature
Table 5 Confusion matrix for Dictionary feature
Fig.9 Graphical representation of Linear Extrapolation of the data with both the Feature set It looks like all these methods have some limitations irrespective of different languages and may come from the technique used to input the data from the annotators.
Fig.9 Graphical representation of Linear Extrapolation of the data with both the Feature set It looks like all these methods have some limitations irrespective of different languages and may come from the technique used to input the data from the annotators.
Related topics:
Computer ScienceDistributed ComputingArtificial IntelligenceNatural Language ProcessingSentiment Analysis
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved