I. INTRODUCTION:

The concept of talking and walking computers are based on NLP and Artificial Intelligence, Here all the AI programs are mostly English due to its globally used for communications, but the developer and scientist are facing issues due to the ambiguity of English pronunciation. The Sanskrit language having very tight grammar and pronunciation making it unambiguous. It was firstly risen by Mr. Rick Briggs a former NASA scientist. The Panini rishi has made 4,000 rules in the book “Ashtadhyayi”[1][2]. The Sanskrit is now like a treasure because it is suitable to use as a language like NLP's need [2].

The first part of this paper is about the NLP and its issue of ambiguity while we use English as a processing language, then the second part will describe the solution for it by using Sanskrit as a processing language for NLP. We will also describe here how the new algorithms will work with AI, NLP and ANN.

a. NLP and its ambiguity:

So, Natural Language Processing has three parts which are respectively three meanings.

Natural:

The thing which is normal for us to use and god gifted or likes by which most of us are aware.

Language:

The method we use to communicate with each other, an interface to communicate. A programming language is used to communicate with Hardware and its gates which are designed to understand with the help of compilers and interpreters.

Processing:

The meaning of processing is to making or modifying a thing to make more suitable or convenient and at last giving a new product.

So, here the actual meaning of NLP is the process of using human languages use by machines like robots through that we can communicate with machines as they are normal human beings [3].

The ambiguity:

The ambiguity in NLP, means the confusion raised during communication between humans and machines and it makes machines unable to extract meaningful and informative data [4]. Because when we talk we take a little pause as per need and the sentence we are saying. As per normal human being can understand with its intelligence which is gained by continuously learning. We don’t like to talk like this “Hey, (there is a pause because of a comma) are you fine? (Question mark)". Ambiguity is raised between sentences we say having the same spellings but the context of it is different. We can say one thing in two ways that is also an ambiguity [5].

b. ARTIFICIAL INTELLIGENCE:

The Intelligence of machines is called artificial or the process of making machines intelligent is AI working.

Artificial:

Which is not real or which is made by humans and work as it is natural.

Intelligence:

The ability of calculation, decision making, identification, and prediction is called intelligence. It is usable for predictions, identifying an object, color or assembly line tasks, human-friendly machines can be friends and can be an employee for us. It is also usable for drawing, the military system, etc.

c. ARTIFICIAL NEURAL NETWORK:

A way of computing systems as the human brain works. The design is a foundation of biological human brains works for learning method. It is used in AI to prove and solve statistical and complex mathematics and predictions problems faster than human brains. It works on algorithms and logical gates.

Issues: -

at ANN there are a lot of sophisticated issues that are left to solve and one of them is how many and how much time a network will take to learn is not defined feasibly. That is called a "global minima and global maxima” point for what a lot of efforts and researches are continually running [6].

d. TALKING COMPUTER:

This word comes from a concept that a machine can make conversation or communication with humans as humans do. Yet to date there are a lot of talking bots are available like "Alexa introduced by AMAZON, Siri developed by Google, and “Cortona by Microsoft ". Also, there is a humanoid robot called "Sophia” which is latest and more advanced than others is invented by Hanson robotics in Hong Kong. It uses Hanson AI mind cloud through all humanoid robots will and is connected with. The concept is that if one robot learns one thing the information will store in a cloud database and the other robots will use it for themselves. We need them for buying something or placing an order in a shop. We can use them as an associate or receptionist and for a telephone-caller person, etc [7].

e. SEMANTIC NETWORKS:

Semantic networks are important constraints for and NLP. These are graphical ways of knowledge representation. In semantic networks, nodes are interconnected to show data patterns. In NLP to solve its ambiguity, the nodes of the semantic network as a flow chart or showcase of connection between words [9]. Like "subject is connected to object with the help of a verb. So, the verb is an agent that helps to accomplish transactions of an action. Like “Ankit is driving a car.”

Figure 1 : Transactions triplets

II. SANSKRIT AS A PROGRAMMING LANGUAGE:

From 1,000 years ago the Sanskrit grammar has created by Panini rishi, and it is unambiguous then other languages. It has words with a different meaning. In Sanskrit, the words are being usable by vibhakti, karaka, sandhi, dhatus. I am not saying that we need to create a new programming language by Sanskrit. The hunter scientist is using the paniniayan grammar to establishing a logic for NLP or data transfer. So this paper is dealing to use Sanskrit as a tool. We know that a programming language is human-readable means the computer can't understand it, for it the linkers, compilers have been invented which converts the high-level languages in low language [8].

III. KNOWLEDGE REPRESENTATION IN SANSKRIT:

a. In English:

There are around 171,476 words in English which respectively make a structure to create sentences. Normally that sentence will confuse to a machine to decide while it is trying to understand what had been said because of human being are emotionally intelligent and aware about the abstract or our brain can conclude (adding situations and sentence to previous one) and that is to us but while machine needs it. It needs the proper input or case scenario for each sentence to be spoken.

For example, let's assume these two sentence: -

1. Gagan gave a bath to his dog wearing a yellow t- shirt.

Ambiguity: -

Is the dog or Gagan who wearing a Yellow t- shirt.

2. I have never tasted a cake quite like that before!

Ambiguity: -

It is not cleared that was the cake good in taste or bad in taste [9].

Now, how the parser will treat these sentences. “Sita gave a Mango to Rama”.

Here are five nodes that can be made and it called triples. Which are labeled in 4 arcs. [10]and [11].

Figure 2 : Triplets

b. Sanskrit words and structure:

The Sanskrit varna vyavstha and dhatus are the relays to work. Here the structure of Sanskrit is like ((Karta, karma Kriya,) (KriyaVishean)). Karta is a Noun or subject; Karma is Object, And Kriya is mean to a verb or the action is happening. In Verb the adverb accomplishes the task to specify the timing and which context it is indicating [11], [12] and [13]. Example:-

“Balakah maatre pustake kalamena kavitam likhati”. Meaning, “A Boy writes a poem for his mother in a book with a pen.”

This Information can be divide as below mentioned triples—

· Write, agent, Boy

· Write, recipient, Mother

· Write, Object, Poem

· Write, Location, Book

· Write, Instrument, Pen

Figure 3 : Tree triplets

IV. Existing system drawbacks:

The existing system is facing issues with language ambiguity problems. The grammar issues and data storage size for any language. There are few languages are accepting new words each year or after a certain incident happens. The word “Selfie” added in the oxford dictionary in 2013 [14]. In the same way, a lot of words are accepted to add in the oxford dictionary means English is not stable. Also English having different variants and accents continental wise[15]. Like:-

· British Accents

· Geordie

· cockney

· Multicultural London English

· Scottish English

· West Country(southwest British)

· Midlands English

· Welsh English

· Estuary English (southeast British).

Here, developers and researchers are getting stuck with problems of “DATASET, TRANSACTION OF DATA, STRUCTURE OF LANGUAGE, EVALUATION and STORAGE CAPACITY.” Humans do not regularly follow the grammar rule while verbal communication they use. The grammar structure in only follow when written communication held. As we know while we talk we may take a pause between each sentence and kind of context we have to share. While a machine reads it say comma, question mark where a person does not say symbols which are reside between words. To teach is this emotional context intelligence. Currently, most languages are not too much suitable.

V. Proposed Idea:

After reviewing researcher’s papers we found Sanskrit is most suitable for the training of a machine. The main thing which makes me strongly agree and convinced me is the structure of Sanskrit grammar and consistency. Our rishi Panini has penned the structure of Sanskrit, and it is not changed since it has been once penned. The uniqueness and strong words each have a meaning. In here the Sanskrit is not targeted to object it just describes the properties of an object [1]. The parsing and parallel computing will fast using Sanskrit.

VI. CONCLUSION:

So, in this paper we discussed NLP basic issues and Why Sanskrit is most suitable for NLP. Also, take care that this paper does not tell that we have use Sanskrit as a programming language or make a new compiler for it. We have to abstract logic which makes it unambiguous than other languages.

The Tight Grammar Structure of this language is still not changed after 1,000 plus years has gone. Currently, very fewer people know Sanskrit, but previously it calls "DEVBHASHA' which meant it is a language in which GOD and GODDESS use this to communicate with each other. In India, all VEDAS and Puranas, Upanishads are written in Sanskrit. In earlier Bhartiya Nivesh each person used it for communication. According to an independent search, the chanting of Sanskrit words helps to relax and make a vocal point and facial muscle power. As all around the Sanskrit is good for NLP as all-around performance, but it is still sophisticated because the grammar is vast and people have understood and learned it before applying it to it.

VII. Further Studies:

It is just a referential study, Later, we will complete it with respected dedication. It is a very good field and yet vast things are rest to developed and find.

VIII. REFERENCES:

1. D. Teja and S. Kothuru, “Sanskrit in Natural Language Processing,” International Journal of Advanced Research in Computer Science and Software Engineering, vol. 5, no. 3, pp. 596–600, 2015. S.

2. S. Saxena and R. Agrawal, "Sanskrit as a Programming language and Natural Language Processing," Global Journal of Management and Business Studies, vol. 3, no. 10, pp. 1135–1142, 2013, Research India publication.

3. A. Bharti, V. Chaitanya, and R. Sangal, pp. 1–212.“Use of Sanskrit for natural language processing,” International Journal of Sanskrit Research, pp. 78–81, 2016.

4. “Vipin Mishra, Sanskrit As Programing Language: possibilities and difficulties," International Journal of Innovative Science, Engineering and Technology, vol. 2, pp. 1089–1096, 2015.

5. Rick Briggs, Knowledge representation in Sanskrit and Artificial Intelligence, RIACS, NASA Ames Research Center, Moffet Field, California 94305, The AI Magzine Volume 6, number1, 1985.

6. C. Bathulapali, D. Desai and M. Kunhere, Use of Sanskrit for natural language processing, International Journal of Sanskrit Research, ISSN: 2394-7519,2016,2(6),78:81

7. P. Saxena, K. Pandey, V. Saxena, Panini’s Grammar in Computer Science, Recent Research in Science and Technology, ISSN:2076-5061, 2011, 3(7), pp. 109-111.

8. K.V. Patel, Sanskrit: Some Insights as a Computer Programming Language, 4^th International Conference on Multidisciplinary Research and Practice, ISBN: 978-93- 5288-448-3, 2017, PP: 179-184.

9. Martine F. R. works as a Technology Journalist at Analytics India Magazine, “How Important are semantic networks in Artificial Intelligence” “https://analyticsindiamag.com/semantic- networks-ai/.

10. Nikhil Malhotra, “Towards an improved man and machine connect using Sanskrit”, {Driving towards natural language and its possibilities in the real world using Sanskrit} march 12-2019- https://medium.com/@nickmalhotra/towards-an-improved-man- and-machine-connect-using-sanskrit-dd6878e20655.

11. Prakhar Shrivastava e.t Team, ACAI 2018: Proceeding of the 2018 “International Conference on Algorithms, Computing and Artificial Intelligence”, December 2018, Article No.:56, PP-1-6, https://doi.org/10.1145/3302425.3302487

12. Sharadha Adinarayan, e.t. Team, “Part-of Speech Tagger For Sanskrit: A State of Art Survey”, International Journal of Applied Engineering Research, 10(9):24173-24169, January 2015. https://www.researchgate.net/publication/277957685_Part- of_Speech_Tagger_For_Sanskrit_A_State_of_Art_Survey.

13. Diego Lopez Yse, “Your Guide to Natural Language Processing (NLP), “How Machine Process and understand human language” January 16-2019, https://towardsdatascience.com/your- guide-to-natural-language-processing-nlp-48ea2511f6e1

14. Selfie https://www.bbc.com/news/uk-24992393

15. Accent https://www.hotcoursesabroad.com/study-in-the- uk/once-you-arrive/the-different-types-of-british-accents/ Date: - 27/08/19 NLP and its ambiguity, AI, ANN, Talking Computer, Sanskrit as a Programming -languages, Sanskrit Grammar

Received on 22.05.2020 Accepted on 20.06.2020

Int. J. Tech. 2020; 10(1):62-66.

DOI: 10.5958/2231-3915.2020.00012.7