Dynamics of Semantic Networks of Independence Day Speeches
0 likes134 views
The document outlines a methodology using semantic linguistic networks combined with natural language processing to analyze Independence Day speeches of Indian Prime Ministers from 2004 to 2014. It demonstrates the dynamic semantic relationships and shifts in policy focus over the years, highlighting the transition from long-term strategic goals to short-term actionable policies. The findings reveal how complex network analysis can assist policymakers in identifying influential terms and emerging semantic patterns in public speeches.
Dynamics of Semantic Networks of Independence Day Speeches
- 1. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Dynamics of Semantic Networks of Independence Day Speeches Prem Sankar C Vinod Chandra S S K. Satheesh Kumar IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 2. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Outline 1 Introduction 2 Proposed Approach 3 Natural Language Processing 4 Complex Networks 5 Results 6 Conclusion 7 Reference IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 3. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Introduction 1 Public speeches of governing people can be an ideal source of information Indication of current trends Expected policies and Long-term plans Pattern shifts in priorities 2 In this work, We proposed a new approach - semantic linguistic networks for identifying semantic patterns from the text corpus. 3 It is a combination of linguistic analysis, semantic knowledge and complex network science concepts. 4 The complex network analysis is relevant to a wide range of systems where individual units interact with each other. 5 It is a combination of graph theory and computer science. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 4. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Methodology 1 Preprocessing the text corpus 2 Construct complex networks of key terms. 3 Network Visualization and Analysis Network Definition Let G(E, V) be the graph derived from the speech text corpus, where V denotes set of key terms (nodes), and E indicates the co-occurrence of these terms in a sentence in the text corpus. As a case study - we collected the text of Independence day speeches of Indian Prime Ministers (2004-2014 period) and and explored the dynamics of semantic relationships in the networks. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 5. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Step1 - Data Preprocessing Tokenization: Every word in the dataset is a treated as a token Normalization: All tokens were converted to lowercase letters. Removing stop words: Stop words are those which do not provide any valuable information and in some cases even reduce the quality of data. Stemming process:is the process of reducing inflected (or sometimes derived) words to their word stem, base or root form. Eg - helped, helping ... to help The major key terms are identified after filtering stop words and stemming process IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 6. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Visualization of Semantic Networks The visualization shows the dynamical changes of the network over the years and reveals the hidden patterns. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 7. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Visualization of Semantic Networks (cont..) If a set of nodes belongs to the similar community (node color), then the semantic correlation between them will be relatively high. Modularity iterative algorithm used for network clustering. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 8. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Visualization of Semantic Networks (cont..) The influential keywords are identified using degree centrality measures and represented with larger node size in the network. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 9. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Network Structural Properties The number edges is increased with time The density of networks increases over time. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 10. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Number of Edges V/s Years 650 700 750 800 850 900 950 2004 2006 2008 2010 2012 2014 NumberofEdges Year The number edges is increased with time IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 11. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Density V/s Years 0.065 0.07 0.075 0.08 0.085 0.09 0.095 2004 2006 2008 2010 2012 2014 NetworkDensity Year The density of networks increases over time. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 12. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Comparison of Networks in 1947 and 2014 In 1947 the ”nation” is attached with work, dream, people, peace and soul. In 2014 the ”nation” is attached with rural, health, child and economics. In 1947, we were more into long-term strategic goals and deals with how to face the world as the new nation, but in 2014 the primary focus is shifted from long-term goals to short terms goals. It mainly deals with how to tackle with current issues and policies to overcome present bottlenecks. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 13. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Comparison of semantic connections of term ’Today’ 1947 - Peace,service, achievement, fortune, and opportunity. 2014 -Child, Village, bank, rural, household, benefit and school Indicates main concerns in (today) present scenario. In 1947, after the independence, the primary focus indicate about the Equal opportunity for all people and provide service to people and national and achieve peace. In 2014 the main connections indicate about providing beneficial services like banking to rural villages, provide education to every child through more schools. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 14. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Major Observations The analysis shows the dynamic changes in the network in the local semantic relationship between the keywords over the years and reveals the hidden conceptual patterns. The most frequently mentioned keywords are identified using degree centrality and represented with larger node size in the network. Helpful in identifying new key terms and its semantic relationships. By comparing the networks of continuous years, we can easily detect the shifts in semantic relationships of keyword e.g., banking. In 2014, the relationship between rural, village and banking shows that policy makers are interested to provide the banking service to the rural villages and later Pradhan Mantri Jan Dhan Jojna - Prime Minister Scheme focused on opening bank accounts for rural and poor people was launched. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 15. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Conclusion Our analysis shows that the complex network of a public text can effectively be utilized to visualize the hidden semantic relationship between different key terms in the corpus. Semantic Linguistic analysis helps policy makers to identify most dominant terms or semantic relationship between terms. Based on the current Indian scenario, the semantic linguistic analysis helps to shed light on the most influential policies in the years to come. It also captures new policy terms and their moment of introduction and traces the change in long-term semantic relationships between the key policy terms. We demonstrated a real world application of the complex network analysis by analyzing and visualizing the text network of Independence day speeches of Indian Prime Ministers. IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 16. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Reference M. Newman, Networks: an introduction. Oxford University Press, 2009 D. Paranyushkin, “Visualization of texts polysingularity using network analysis,” Prototype Letters, vol. 2, no. 3, pp. 256–278, 2011. M. E. Newman, “Fast algorithm for detecting community structure in networks,” Physical review E, vol. 69, no. 6, p. 066133, 2004. S. Scott and S. Matwin, “Feature engineering for text classification,” in ICML, vol. 99, 1999, pp. 379–388 IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala
- 17. Introduction Proposed Approach Natural Language Processing Complex Networks Results Conclusion Reference Thank You. . . . Acknowledgement The authors wish to express their gratitude to the Campus Computing Facility at University of Kerala set up under DST purse programme for providing their facilities. Presented By Prem Sankar C Research Scholar Department of Futures Studies University of Kerala- India IC4 2018 - The International Conference on Control, Communication and Computing Dept. of Futures Studies,University of Kerala