– The requirement is not only to report the results but to generate them. Generating the results requires a certain set of highly technical skills in the field of Computer Science in general and Artificial Intelligence in specific. As a very quick overview, this includes data collection, data preprocessing, AI/ML model development (in Python), results analysis, etc.
– The dissertation deals with Big Data, somewhere around 1 million tweets with heavy Deep Learning analysis using Python algorithms.
– Along with the document submission, every used material should be attached, including Python codes files, datasets (csv), and any other applicable materials.
– I have to be able to run the code and obtain the reported results (as a demo) during my defense.
– Some quick numbers regarding some of the requirements: 80k words (excluding appendences, tables, diagrams, and references), proper citations using the BUiD Harvard referencing guide, etc.
– The paper structure/chapters according the university’s guidelines and standards (to be shared later).
Some of my requirements with reference to PaperHelp form:
– 80k words.
– References (highly academic and cited) and graphs.
– Top writer(s); has to be the same team throughout the research for obvious reasons.
The pricing for all of the above requirements came out considerably high which I’m willing to commit if everything was agreed upon. However, a split payment, such as 50% upfront & 50% on final version or something similar, would be ideal.
Naturally, the submission will be heavily reviewed and discussed with my supervisors and examiners for comments/revisions/corrections/etc., therefore, I will need way more time for revision deadlines. In fact, one of your representative once told me that for such huge projects, there kind-of no deadlines on revisions, which is great. Nevertheless, I will try my best to provide a tentative as we move forward but I cannot guarantee that I will always be able to because there are many factors into this. One last thing, please note that the project duration will at least be around 6-12 months, including revisions, comments, and back-and-forth discussions.
Finally, below is my proposal abstract:
Various sectors all over the world have been significantly benefiting from and extensively utilizing the concept of sentiment analysis to closely study and examine public opinions and users’ experiences. In addition, the emergence of social networking and its various platforms has tremendously enhanced the abilities of opinion mining and enabled enterprises to move even closer to their customers, viewers, and followers. Nevertheless, textual posts and messages have considerable limitations within their nature due to their language restraints, in addition to certain difficulties in conveying rather sophisticated emotions or expressions. Emojis, on the other hand, have been heavily invested in and introduced by major technology companies to address these specific textual contents’ limitations and boundaries. With simple, convenient, yet tiny icons, emojis can convey highly sophisticated emotions and sentiments efficiently regardless of the user’s native language. It has been evidently observed in the literature that the majority of the former research in this area pursue a certain pattern by creating or adopting emojis lexicons and try to implement a well-trained machine learning classifier to enhance the annotations’ accuracy. Although many decent outcomes have been generated, this approach seems incapable of handling the complication of emojis’ multiple meanings and implications such as sarcasm and humor. The proposed Ph.D. thesis aims to establish an original implementation for the state-of-the-art sentiment analysis models with a unique emphasis on emojis. The study will analyze people’s sentiments toward the emerging regulations and restrictions during the COVID-19 pandemic in the UAE and other adjacent regions to aid disaster management authorities in their subsequent announcements. The proposed research will adopt a quantitative approach by applying the conventional yet most effective sentiment analysis framework composing of data acquisition and preparation, model development and training, and validation and evaluation. In particular, the study will utilize two of the most distinguishably achieving models in terms of sentiment analysis accuracy: LSTM and Multinomial Naïve Bayes classifiers using Twitter’s data. Additionally, this proposal highlighted the most pressing ethical obligations concerning analyses with public data involvement and empirical studies in general and the undeniable importance of adhering to such policies and considerations. Finally, some potential technical as well as logistic limitations and constraints were addressed concerning natural language processing prominent imperfections, languages and regions restrictions, and disregarded techniques and approaches.
