Building a Swahili NLP hub
A Swahili NLP hub for everyone - That is developers, data scientists and everyone with no-coding experience.
ABOUT US
Let us share with you a little bit of history.
Fonimu was founded by a group of enthusiasts who believe NLP can revolutionize access to services in Africa. This has been the core of our drive. Today, we are a community solving the challenging Swahili NLP problems that will impact millions of people across Africa, as well as look toward the future.
What we want to accomplish
We’re trying to build a Swahili NLP hub, and we're doing it with the hope of having a huge, positive impact in Africa. That means at Fonimu, you'll find APIs for various Swahili NLP tasks, pretrained models, links to Swahili datasets and you'll have an opportunity to make your contributions. We want Fonimu to be the easiest Swahili NLP ecosystem for developers, data scientists and everyone with no-coding experience.
Problems we are solving
The Swahili NLP problems we work on are hard, and we want as many of the great minds thinking about them as possible. Below are the problems we are working on;
Text Summarization
The extensive generation of data and content on the internet has given rise to a need for text summarization to generate efficiency and reduce reading time. This is well demonstrated on Online magazines, research sites, and news aggregator apps that rely on short, informative summaries to keep readers interested and up to date.
Text Classification
Text can be an extremely rich source of information, but extracting insights from it can be hard and time-consuming, due to its unstructured nature. WIth text classification a set of predefined categories can be assigned to an open-ended text. Some examples of text classification are email classification, news classification and sentiment analysis.
Named Entity Recognition
Named entity recognition (NER) is being used heavily in the industries today to automate major categorization of data for unstructured text and datasets. NER seeks to locate and classify named entities in text into predefined categories such as the names of persons, organizations, locations, etc.Human resources: Speed up the hiring process by summarizing applicants' CVs
Grammar Correction Tool
We often encounter grammatical errors when writing an email or an imperative document. This usually causes us to spend a large amount of effort and time on finding and correcting the grammar error. Grammar Correction Tools help in minimizing mistakes in spelling, punctuation, grammar, and word choice—an area known broadly in natural language processing (NLP) research as grammatical error correction.
Text Generation
Text generation helps in writing/producing high-quality content quickly and efficiently. Generating text provides a simple step to ensure that you can create something with as little effort as possible. Examples of text generation are; Translation, Text autocompletion, Question Answering, Code generation.
Demo
Fonimu Use Cases
We are fostering some of the great work in African NLP especially in Swahili. We Try some of the demos of the use cases that we have been working on so far.
Mask filling
Mask token: [ _ ]
Computed result
mwaka
9%
kenya
6%
historia
4%
ya
4%
dunia
3%
Named Entity Recognition
Identify specific entities in your swahili sentence
Computed result
Text autocompletion
Swahili text auto completion demo
Computed result (select any suggestion)
Sentiment Analysis
Check if a sentence is positive, negative or neutral.
Computed result
positive
70%
negative
6%
neutral
24%
News classification
categorize a news content into respective topic
Computed result
uchumi-economy
0%
kitaifa-national
93%
michezo-sports
1%
kimataifa-international
4%
burudani-entertainment
0%
afya-health
0%
Datasets
Swahili Datasets
Datasets play a crucial role in developing accurate models. The quality and quantity of the data used impacts the model's performance. Fonimu API leverages the use of annotated corpora, question-answering datasets, and sentiment analysis datasets.
With the availability of Fonimu datasets, researchers and developers can create new models in swahili
Frequently Asked Questions
How does a Fonimu work?
Is Fonimu Free?
Reprehenderit tenetur?
Honsectetur adipisicing?
What are the features of Fonimu
How does a Fonimu work?
Lorem ipsum dolor sit amet?
Amet asperiores at?
