A Chatbot for Answering Frequently Asked Questions by University Students Using Natural Language Processing and Multinomial Naïvebayes Algorithm

1Ede Ifesinachi Chizzy; 2Aminu Muhd Bui & 3Hassan Suru
1Department of Computer, Federal University, Bennin Kebbi
2Department of Computer, Usman Danfodio University, Sokoto
3Department of Computer, Kebbi State University of Science and Technology, Aliero
Email: edechizzy@yahoo.co.uk

ABSTRACT

Chatbots are programs that impersonate human discussion and their plan should be possible utilizing different techniques. Be that as it may, little work has been finished in the use of chatbots in the instructive area, thus; this undertaking is centered on making a chatbot to be utilized by understudies to respond to their habitually posed inquiries from the school’s web-based entertainment stage and regulatory workplaces. This KSUSTA Chatbot has the ability to make discussions; answer the course and workforce subtleties; answer the regularly posed inquiries instead of looking at a considerable rundown of FAQ’s searching for replies. Issues engaged with making Chatbots are information assortment which winds up giving not exactly needed measure of information for preparing and retraining, utilization of Programming interface’s which decreases adaptability in the bot. Be that as it may, these issues were handled by involving BeautifulSoup for information assortment, Pandas for information handling and Multinomial
Naïve Bayes model with a superior presentation. To develop the Chatbot, Python Language was utilized as the fundamental language. Furthermore, its AI and Natural Language Processing Libraries, web scrapping and document handling instruments were utilized, the front end graphical UI (GUI) was designed utilizing Flask (python), Html, CSS and JavaScript, Data set was
taken care of with PostgreSQL, for recovery and retraining. The system has achieved 84% accuracy of correctly classifying the questions. One of the major drawbacks was the imbalanced state of the data set. Below are the various metrics that were used to evaluate its performance. Keywords— Frequently Asked Questions (FAQs), Natural language processing (NLP), Machine learning (ML), Multinomial Naive Bayes, Supervised Learning, Reinforced Learning

Comparative Analysis of Gum Arabic and Molasses (Binders) in Briquette Produced from Millet Husk

*1Arzika Abdullahi Tambuwal, 2Abdulhameed Salihu, 1Yunusa
Muhammad Boyi, 3Sani Garba, 1Adamu Salisu Muhammad, 3Yusuf Sahabi, 3Hauwa’u Umar and 3Mubarak Muhammad.

1Department of Chemistry, Shehu Shagari College of Education, Sokoto

2Department of Physical Sciences, Niger State Polytechnic, Zungeru, Niger

3Department of Science Laboratory Technology, Umaru Ali Shinkafi Polytechnic Sokoto
*E-mail:arzikatambuwal1982@gmail.com

ABSTRACT

The study was carried out to investigate the effects of binders (molasses and gum Arabic) on millet husk in the production of briquettes. Fixed quantities of millet husk were used to produce briquettes with varying percentage of binders (10%, 20% and 30%). Low pressure fabricated briquetting machine was used for compression to produce the briquettes, after sun drying to reduce the moisture content to minimum value. The
proximate analysis conducted, indicated the range of moisture content% (2.1-3.0) ash content% (7.8-11.4) volatile matter% (61.9-76.6) and fixed carbon% (13.0-26.5). The physical properties had the values ranging from (0.52-0.60), (0.18-0.24) (1.69-1.80), (2.31-3.14), (3.5-23.2), (4.7-30.2) for compressed density (g/cm3), relaxed density (g/cm3), compaction ratio, relaxation ratio, durability (%) and water resistant (sec) respectively. The
fuel density included ignition time (sec), after glow (sec), boiling time (mins) and calorific value (KJ/kg) with the value ranging from (2.3-8.3), (2.0-24.4), (16.10-19.13) and (29830.95-30119.84) respectively. The study shows that millet husk with gum Arabic serves as a better combination for the production of briquettes.
Keywords: Millet husk, Gum arabic, Molasses.