Sentiment analysis is basically opinion mining or emotion analysis. Many people express their views and sentiments through verbal, non-verbal and written forms to show their opinions and emotions on products, personalities, tourist places, educational institutions, hospitals, historical places, government, restaurants etc. A number of organizations are planning and concentrating on views and opinions of people to get some useful information. The social media, public and private sector organizations websites, web pages, blogs and online surveys are the important sources for getting opinions and reviews of people, thus, word wide web is best source of generating such types of data. Sentiment analysis, review analysis, emotion detection and opinion mining are procedures of analysing the unstructured or structured data for the purpose of evaluation of sentiments and opinions. Sentiments show the scale or level of confidence for positive opinion, negative opinion or neutral opinion or sentiments. Today, sentiments and opinions or reviews evaluation are one of the significant attentions of Natural Languages Processing generally called NLP. Majority of computational linguistics and sentiment analysis etc. software applications are existing for English and some other languages, nonetheless, numerous languages are there which cannot meet the level and category of these types of languages. Though, research studies and tools development processes are in growth for the languages, which are not resourced languages yet. The Sindhi language is an Asian language, which may be called the morphologically rich language, nevertheless, it faces several complexities since evaluating and analysing the online or offline text. Though, lots of data are available online or offline in different forms but yet no appropriate research study or work has discovered in the field of NLP as well as on sentiment analysis for Sindhi language text particularly. The deficiency of development work and research studies as well as technical resources for Sindhi language make the current research work or study interesting and challenging. Viewing and assessing this challenge, we have taken this task to work more to address the problems of Sindhi language data. Therefore, we have focused the construction of text corpus, data set, sentiment analysis system, word tokenization, part of speech tagging as well as subjective lexicon assessment for Sindhi language text. Supporting tools such as Sindhi POS tagger helps in identifying sentiments from Sindhi text corpus. This study has developed the NLP resources including sentiment analysis resources for Sindhi language text. Separate text corpus and linguistic data sets are developed and analysed by machine learning and deep learning models. Machine learning models are trained with small sentiment-based Sindhi training data and large sentiment-based Sindhi training data. The results confirm the proper performance and execution of supervised machine learning models in form of extraction of appropriate sentiments. The sentiment analysis for Sindhi text is done on document-level sentiment analysis, product level and aspect level sentiment analysis. The leaning model is designed and developed for the purpose of sentiment evaluation and analysis for Sindhi language text. Neural network based LSTM model is used with multiple layers to evaluate and validate the sentiment based Sindhi language text and products feature based data set. Results of models confirm the significance of methodology by showing good sentiment analysis and opinion analysis on Sindhi language text. Research study contributes the Sindhi language plain text corpus, linguistics dataset, aspect-based sentiment analysis dataset to the fields of natural languages processing as well as computational linguistics. Sentiment analysis system, which is developed for the Sindhi text is significant and state-ofthe art work. The work places the Sindhi language for international research to explore the grammatical and morphological complexities, perform the information retrieving, language modelling, semantic and sentiment analysis, universal dependencies and unsupervised modelling for text analysis etc.