ABSTRACT
Natural language processing (NLP) is incomplete without graphs and its topologies. NLP includes text, text construction or formation, speech or conversation, grammar rules, Machine interpretation and extraction of the knowledge. The topologies and algorithms of graphs have played a vital role to model the specified problems and the attractive solutions. Topologies of graphs are utilized in NLP in numerous applications like finding objects that fulfill certain primary properties unmistakable concerning different elements, finding internationally ideal answers for the given relations between elements.
Construction of graph topology is a crucial point in NLP as the architecture must model the data appropriately so that it will not exclusively tackle the objective NLP issue, however to solve it in a better way of computation. Graph representation plays an important role to solve the issues of NLP in better computational manners with reference to time and memory space, Due to this, a critical analysis of graph topologies is required according to NLP issues like text structure, discourse, generation, normalization and summarization, syntactic parsing, tagging, machine translation etc. In this chapter the analysis of some graph topologies or architecture is described according to the NLP issues like heterogeneous graphs, multi-layered graphs, hyper graphs. More complicated issues of graph embeddings or data and supported computational approaches are also discussed.
Numerous calculations depend on NP-hard graph classes which don't scale to introduce data sizes. Adaptability is a critical component for these types of calculations, as they need to deal with an ever increasing amount of information. There are additional issues that relate precisely to computational techniques in NLP. For instance, streaming graphs because these types of graphs change on schedule. The graphs developed from web-based media are instances of this. Organizations address the clients and their tweets and the relations between them change rapidly.
There are some open source NLP libraries and tools. These libraries give the algorithmic structure squares of NLP in the applications of real world like Apache OpenNLP, MALLET, TextBlob, CogCompNLP, PyTorch-NLP, SpaCy, OpenNLP, StanfordNLP, etc. for the various applications of NLP like word sense induction and word sense disambiguation. Sentiment analysis and social networks, machine translation, information extraction/knowledge extraction and representation/events, automated documents of clinical documents, etc.
