Bottom up Fuzzy Parsers: Fuzzy Simple LR, Fuzzy Canonical LR and Fuzzy LALR Parsers for Parsing Natural language

- Humans convey the information through natural language. It is a prevailing tool used by peoples in daily life. Natural Language Processing (NLP) involves methods for analyzing the words through many levels of linguistic analysis. Language exhibits mainly two functionalities. First functionality stipulates syntax and second functionality specifies semantics of the language. The development of Fuzzy parsers for performing syntax analysis of the Natural Language (NL) is described in this paper. Conventional Bottom up parsing algorithms such as Simple Left to Right (SLR), Canonical Left to Right(CLR) and Look Ahead Left to Right(LALR) parsers are enhanced by applying Fuzzy Logic (FL). Left to Right (LR) syntax analysis technique is a constructive method for parsing context free languages.


II.
METHODOLOGY Conventional parsing methods are Enhances by fuzzy constructs. Here English language sentence is given as an input and syntactic correctness is tested using fuzzy bottom up parsers. Fuzzy parser model is represented in Fig.1 .This section describes the construction of fuzzy context free grammar for the parsers.

A. Fuzzy Context Free Grammar (FCFG):
Consider commonly used Production rules for construction of English language sentences are as follows [1], [3]   If any contradictory actions occur from the above rules, then the grammar is not LR (1) . The algorithm fails to produce a parser for this case.

The initial state of the parser is constructed from set of items containing [S' .S, $] D. Constructing an FLALR Parsing table [2]
An augmented grammar G' is considered as an input. FLALR Parsing table functions ACTION and GOTO are computed for G'. METHOD: 1. Construct collection of item sets S = (P0,P1,………..Pn) of LR(1) Items for G' . 2. For each core present among the set of LR (1) items, find all sets having that core, and replace these sets by their union.
3. Let S' = {Q0, Q1, …..Qm} be the resulting sets of LR (1) items. The parsing actions for state i are determined from Ji in the same approach as in Canonical LR algorithm. If there is a parsing action conflict, the algorithm fails to produce a parser, and the grammar is said not to be an LALR (1).
4. The GOTO table is constructed as follows. If J is the union of one or many sets of LR (1) items, that is, J = P1 ∩ P2 ∩ ………∩ Pk , then the cores of GOTO (P1,X ), GOTO(P2, X), ………..,GOTO(Pk, X) are the same, since P1,P2,….,Pk all have the same core. Let K be the union of all sets of items having the same core as GOTO (P1,X). Then GOTO (Q,X) = K .

III.
RESULT ANALYSIS The following results shows the input sentence and permutations generated and also shows the degree of fuzziness for the parsed input. Finally it shows the completely parsed sentence with maximum fuzziness . Fig 2. Shows the input given for the FSLR parser.  Here English sentence input is parsed using Fuzzy Canonical Left to Right (FCLR) parser. In this approach initially the grammar rules are defined and action and go to table is constructed, using this table parsing is done for the given input sentence. Result has been explained in Fig 4. Shows completely parsed sentence with its associated fuzzy value.   Right (FLALR) parsers are discussed in this paper. Here FCFG is designed for parsing Natural language. These algorithms are implemented in 'C' Programming Language. Considering English language sentence as an input, permutations are generated. For the generated permutations these Fuzzy LR algorithms are applied. Finally Fuzzy max-min technique is applied to get the degree of fuzziness. Results are discussed in this paper. Our research work involves design and implementation of fuzzy parsers over conventional parser is that it gives degree of fuzziness and syntactic correctness for partially parsed sentences but in conventional parsers the sentences parsed completely are only accepted and rejected completely if it is partially parsed. Syntax analysis assists to get better identification rates significantly. We conclude that among these algorithms FLALR gives improved result and minimizes the states compare to FCLR. Main limitation of FCLR algorithm is number of states generated are more compare to FSLR algorithm.