26
Detection of Semantic Errors from Simple Bangla Sentences Authors K. M. Azharul Hasan, Muhammad Hozaifa, Sanjoy Dutta Presented By Muhammad Hozaifa Department of Computer Science and Engineering (CSE) Khulna University of Engineering & Technology Khulna 9203, Bangladesh. 1 ICCIT 2014

Detection of semantic errors from simple bangla sentences

Embed Size (px)

Citation preview

1. Detection of Semantic Errors from Simple Bangla Sentences Authors K. M. Azharul Hasan, Muhammad Hozaifa, Sanjoy Dutta Presented By Muhammad Hozaifa Department of Computer Science and Engineering (CSE) Khulna University of Engineering & Technology Khulna 9203, Bangladesh. 1ICCIT 2014 2. CONTENTS Problem Statement Objectives & Goal Semantic Error Detection from Simple Bangla Sentences Implementations Experimental results Conclusion and future works 2ICCIT 2014 3. PROBLEM STATEMENT Semantics expresses a range of ideas, from the language to the highly technical. Used in ordinary language for denoting a problem of understanding In linguistics, semantics deals with the study of meaning, changes in meaning, and the principles that govern the relationship between sentences or words and their meanings The research on semantic correctness checking is very important for the purpose of machine learning, opinion mining and intelligent agent development for human computer interaction. 3ICCIT 2014 4. PROBLEM STATEMENT 4 For Simple sentence Semantically Wrong Semantically Correct ICCIT 2014 5. CONTENTS Problem Statement Related Works Objectives & Goal Semantic Error Detection from Simple Bangla Sentences Implementations Experimental results Conclusion and future works 5ICCIT 2014 6. RELATED WORKS Soma Paul describes an analysis of the unification two verb Bangla sentences (V1 and V2) by using semantic principle of compounding Based on HPSG structure. Beth Levin discovers the behavior of a verb, particularly with respect to the expression and interpretation of its arguments, Levin classifies over 3,000 English verbs according to shared meaning and behavior into different categories. ICCIT 2014 6 7. CONTENTS Problem Statement Objectives & Goal Semantic Error Detection from Simple Bangla Sentences Implementations Experimental results Conclusion and future works 7ICCIT 2014 8. OBJECTIVES & GOAL Check the semantic correctness. Establishment of Subject Verb (SV) relationship and Object Verb (OV) relationship to check the semantic correctness of both SV and OV relation. 8ICCIT 2014 9. CONTENTS Problem Statement Objectives & Goal Semantic Error Detection from Simple Bangla Sentences Implementations Experimental results Conclusion and future works 9ICCIT 2014 10. SEMANTIC ERROR DETECTION FROM SIMPLE BANGLA SENTENCES For this work, we follow Simple Sentence format of Bangla text. Subject +Object + Verb (SOV) In the SOV form, the relation of the verb with subject and object is of two fold: 1. Whether the verb with the subject has a well formed structure with semantic compatibility (SV relation) 2. Whether the Object and Verb (OV relation) has semantic compatibility. For explanation, we consider some semantically wrong examples, 1. (SV error) 2. (OV error) 10ICCIT 2014 11. SEMANTIC ERROR DETECTION FROM SIMPLE BANGLA SENTENCES SV relationship also depends on person and tense. 11ICCIT 2014 12. SEMANTIC ERROR DETECTION FROM SIMPLE BANGLA SENTENCES Steps towards solution: 1. Noun Categorization I. Animal Approval Category II. Clipping-Animal approval category 2. Verb Categorization I. Main-Verb Category II. Verb-with-Inflection Category 3. Relationship Validation and Acceptance Checking 4. Take decision whether the sentence is correct or wrong. 12ICCIT 2014 13. SEMANTIC ERROR DETECTION FROM SIMPLE BANGLA SENTENCES 13ICCIT 2014 14. SEMANTIC ERROR DETECTION FROM SIMPLE BANGLA SENTENCES Verb 1 2 (VF) 2 (F) 3 (F) 2/3 (P) bla ami boli tui bolish tumi blo she ble apni blen khola ami khuli tui khulish tumi kholo she khole apni kholen khla ami kheli tui khelish tumi khlo she khle apni khlen chena ami chini tui chinish tumi cheno she chene apni chenen jana ami jani tui janish tumi jano she jane apni janen hoa ami hoi tui hosh tumi ho she he apni hn dhoa ami dhui tui dhush tumi dhoo she dhoe apni dhon khaoa ami khai tui khash tumi khao she khae apni khan deoa ami dii tui dish tumi dao she de apni dn 14 Table : Forms of Simple present tense ICCIT 2014 15. CONTENTS Problem Statement Objectives & Goal Semantic Error Detection from Simple Bangla Sentences Implementations Experimental results Conclusion and future works 15ICCIT 2014 16. IMPLEMENTATIONS Developed a corpus of our own in which we used almost 10000 words in different categories. In this corpus, we tried to sort all types of verbs of Bangla language. Verb Categories : 57 types Subject Categories : 50 types So in our corpus, we tried to cover almost 2850 categories to find out the relations. ICCIT 2014 16 17. IMPLEMENTATIONS Strategy of Semantic analysis (a) Take sentences and check formation with word (b) Check the subject (c) Check the verb (d) Check the relationship between Subject and Verb with proper Object. (e) Take a decision for semantic formulation 17ICCIT 2014 18. IMPLEMENTATIONS (a) Take sentences and check for SOV standard sentence. (b) Check the subject 1. Animal Approval Category 2. Clipping-Animal approval category Subject category - : ;(Second person) Sample sentence: Clipped process: (Clipped) => => 18ICCIT 2014 19. IMPLEMENTATIONS (c) Check the verb 1. Main-Verb Category 2. Verb-with-Inflection Category 19 Sentence Verb Decision + (Verb-inflection) Verb : Category : + (Verb Inflection) Verb : Category : ICCIT 2014 20. IMPLEMENTATIONS (d) Check the relationship between Subject and Verb with proper Object. 20 Sentence : Subject : Verb : Object :( ^ ^ True Sentence : Subject : Verb : Object : ( ^ ^ False ICCIT 2014 21. IMPLEMENTATIONS (e) Take a decision for semantic formulation 1.If the SVO relationship is okay. 2. If the S-V relationship is okay but SVO relationship is not okay. 3. If the S-V relationship is not okay. 21ICCIT 2014 22. CONTENTS Problem Statement Objectives & Goal Semantic Error Detection from Simple Bangla Sentences Implementations Experimental results Conclusion and future works 22ICCIT 2014 23. EXPERIMENTAL RESULTS SOV standard = 05 & Others = 03 Error = 01 Correct=04 Not detected=03 23ICCIT 2014 24. CONTENTS Problem Statement Objectives & Goal Semantic Error Detection from Simple Bangla Sentences Implementations Experimental results Conclusion and future works 24ICCIT 2014 25. CONCLUSION AND FUTURE WORKS In this paper, we have worked on to detect the semantic analysis for SIMPLE Bangla text. Important and Necessary to complete the VALIDATION TABLE and OBJECT VERB RELATION TABLE for all the verbs and nouns of Bangla language. There are a LOT OF SCOPES in this field. We only worked on SIMPLE SENTENCE (SOV). Although there are also COMPLEX and COMPOUND structures in Bangla grammar. This PROPOSED ALGORITHM can easily be extended for complex and compound sentences for semantic error detection. 25ICCIT 2014 26. Thank You . 26ICCIT 2014