https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
428
AI-POWERED SYSTEM FOR FILTERING RELIGIOUS EDUCATIONAL
CONTENT FOR YOUTH IN THE DIGITAL SPACE
To’rabekova Shirin Khaitvoy qizi
Uzbekistan International Islamic Academy
1st-year student of Information Security
Abstract:
In the current era of widespread digitalization, young people
increasingly turn to the internet to seek information about religious teachings.
However, the lack of content regulation has made it easier for youth to be exposed
to misinformation, extremist ideologies, or distorted interpretations. This paper
proposes an AI-based system for filtering religious educational content aimed at
protecting youth in the digital environment. The study explores machine learning
algorithms, natural language processing (NLP) techniques, and ethical
considerations necessary to ensure that the filtered content is accurate, culturally
appropriate, and pedagogically sound.
Keywords:
Artificial intelligence, content filtering, religious education,
digital safety, youth, natural language processing, cyber ethics
The rapid growth of digital media has transformed the way religious
knowledge is disseminated, especially among young internet users. While this
trend presents an opportunity for broad access to spiritual education, it also raises
concerns regarding content authenticity, ideological bias, and psychological
impact. Misleading religious content
—
whether accidental or deliberate
—
can
negatively influence impressionable minds.
Given this challenge, there is a growing demand for intelligent systems that
can automatically identify and filter online religious educational material.
Traditional web filtering mechanisms are often rule-based and insufficiently
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
429
nuanced to differentiate between authentic, scholarly religious content and
distorted, harmful information. Artificial Intelligence (AI), particularly Natural
Language Processing (NLP), offers a promising solution to this issue by enabling
systems to understand context, intent, and semantic structure of online texts.
This paper proposes the design of an AI-powered content filtering
framework tailored for religious educational materials targeted at youth, ensuring
both safety and educational benefit in the digital space.
The uncontrolled flow of digital information presents a double-edged sword,
especially in the domain of religious education. On the one hand, youth now have
unprecedented access to sermons, scriptures, and commentaries across different
schools of thought. On the other hand, this same access leaves them vulnerable to
exposure to
non-accredited, ideologically extreme, or deliberately misleading
content
, often masked as authentic religious material.
Existing content moderation systems, such as keyword filtering and manual
review, are often
inadequate
, as they fail to capture
theological nuance
,
semantic
intent
, and
emotional tone
—
all critical factors in religious discourse.
Furthermore, many social media platforms and video-sharing sites lack
regionally
contextualized content governance
, making young users in Muslim-majority
societies like Uzbekistan or Indonesia particularly vulnerable to imported radical
narratives.
Artificial Intelligence (AI), particularly with recent advances in
transformer-based language models
and
context-aware classification systems
,
now provides a promising solution to this issue. Through
machine learning (ML)
and
natural language processing (NLP)
techniques, AI systems can be trained to
distinguish between legitimate, scholarly Islamic discourse and content that
misrepresents or manipulates religious principles for ideological gain.
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
430
This research aims to address the following core questions:
•
How can AI systems accurately detect and classify religious educational
content?
•
What metrics and ethical frameworks are needed to balance
freedom of
belief
with
content integrity
?
•
Can AI adapt to diverse Islamic perspectives while still filtering harmful
content?
By focusing on the development of an
AI-powered filtering system
, this paper
seeks to support
youth protection
,
religious integrity
, and
digital literacy
in an
age where online content increasingly shapes belief systems and identity formation.
The development of the proposed filtering system involves several key
stages:
2.1 Dataset Collection
A diverse corpus of religious educational texts (Quranic exegesis, Hadith
literature, sermons, and modern interpretations) was collected from authenticated
scholarly sources and online platforms. A secondary dataset includes texts flagged
for containing distorted or extremist ideologies.
2.2 Preprocessing and Labeling
Texts were cleaned, tokenized, and annotated by a panel of religious scholars and
linguists. Labels were assigned into categories such as:
•
Authentic
(verified scholarly content),
•
Moderate
(general religious commentary),
•
Problematic
(misinterpretation, incitement, extremism).
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
431
2.3 Model Architecture
We employed a hybrid NLP model consisting of:
•
BERT-based language model
for contextual understanding;
•
CNN-LSTM architecture
for classification;
•
Sentiment and toxicity analysis
modules for emotional tone evaluation.
The model was trained to classify content based on theological correctness,
semantic clarity, and emotional neutrality.
2.4 Evaluation Metrics
Performance was measured using:
•
Accuracy
,
•
Precision
,
•
Recall
,
•
F1-score
,
•
False positive rate
, especially regarding sensitive or ambiguous texts.
The AI system was tested on a validation set of 5,000 texts, yielding the following
results:
Metric
Value
Accuracy
92.8%
Precision (Authentic)
94.1%
Recall (Problematic)
89.3%
F1-Score
91.2%
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
432
Metric
Value
False Positives
3.7%
The system effectively filtered out flagged content while preserving access
to authentic educational resources. An interface prototype was also developed for
integration into web browsers and mobile apps.
The study demonstrates that AI can serve as a powerful tool for safeguarding
religious education in digital environments. Compared to manual filtering or rule-
based algorithms, the AI system showed higher adaptability to contextual and
semantic nuances.
However, several challenges remain:
•
Contextual ambiguity:
Some texts require deep theological interpretation,
beyond the reach of current AI capabilities.
•
Multilingual limitations:
The system was primarily trained on English and
Uzbek; further work is needed to support Arabic and other regional
languages.
•
Ethical concerns:
Filtering must respect freedom of belief while upholding
national and religious values.
Close collaboration with religious scholars, ethicists, and educators is vital to
maintain the integrity of AI filtering in this sensitive domain.
The findings of this study highlight the potential of AI-based systems in
addressing one of the most pressing challenges of our time: safeguarding young
people from misinformation and harmful narratives within digital religious content.
The high accuracy of the AI model in distinguishing authentic from problematic
content shows that
machine learning techniques
—
when properly trained and
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
433
ethically guided
—
can become powerful tools for content governance
in
religious education.
One of the key strengths of the system lies in its
context-awareness
. Unlike
traditional keyword-based filters, the NLP-driven approach was able to assess the
semantic intent
of a passage, including subtle nuances of interpretation, emotional
tone, and theological framing. This is especially critical in religious discourse,
where meaning often depends on context, tone, and scholarly consensus.
However, several limitations and concerns must be acknowledged:
Ethical Concerns
Implementing an AI system to filter religious content must be approached
with
extreme sensitivity
. Who decides what is “authentic” or “extreme”? There is
no universal agreement across all schools of Islamic thought. Therefore, such
systems must be guided by a
pluralistic and consultative framework
, involving
recognized scholars and educators from different traditions.
Risk of Overblocking
While the AI system performed well, there were still instances of
false
positives
, where content was mistakenly flagged due to strong language or
complex interpretations. Overblocking can
limit access to valuable knowledge
and stifle critical engagement with diverse perspectives.
Adaptability and Localization
Another major challenge is adapting the system for
multiple languages and
cultural contexts
. Religious expressions in Arabic, English, Uzbek, or Malay may
vary in terminology, rhetorical style, or legal emphasis. Localization of the AI
system is essential to maintain its effectiveness and cultural relevance.
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
434
Need for Human Oversight
Despite automation,
human moderation remains essential
, especially for
edge cases where AI lacks the nuance to make sound judgments. A hybrid model
—
combining AI-driven suggestions with expert review
—
offers the most balanced
solution.
Comparisons with Other Studies
Our approach aligns with similar efforts in content moderation (Zannettou et
al., 2020), hate speech detection (Khan & Rehman, 2020), and digital religious
literacy frameworks. However, this study is distinct in focusing specifically on
religious education for youth
and in proposing a
structured ethical review
mechanism
within the AI pipeline.
In sum, the system represents a promising step toward a
safe, informed, and
inclusive digital environment
for religious learning. However, it must evolve
continuously to remain responsive to theological complexity, user needs, and
technological change.
This research presents a practical approach to designing an AI-powered
system capable of filtering religious educational content for youth in digital spaces.
The system has proven effective in identifying and blocking problematic content
while supporting access to verified teachings.
Future developments may focus on:
•
Expanding multilingual capabilities;
•
Incorporating video and multimedia content analysis;
•
Building adaptive filters based on age, sect, or regional preferences.
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
435
Ultimately, this initiative contributes to safer digital spiritual education and
promotes responsible AI use in culturally and morally sensitive contexts.
This research underscores the crucial role that artificial intelligence can play
in moderating religious educational content online, particularly for vulnerable
groups such as youth. By combining natural language processing with ethical
oversight, the proposed AI-powered filtering system achieved promising results in
identifying, classifying, and blocking problematic content while preserving access
to verified, scholarly material.
However, the deployment of such a system must be rooted in
transparency,
inclusivity, and religious legitimacy
. The system cannot function in isolation
from
human scholars, ethicists, and educators
, whose insights are necessary to
refine its judgment and ensure theological neutrality. In this respect, our framework
lays the groundwork for a
collaborative model
—
where AI handles large-scale
filtering, and human expertise ensures doctrinal soundness and ethical alignment.
Future work will focus on:
•
Expanding language support
, especially for Arabic, Turkish, Persian, and
regional dialects;
•
Incorporating multimedia filtering
, including videos and audio content
that often carry unfiltered ideological messaging;
•
Enhancing user feedback loops
, allowing users and scholars to challenge
or refine AI decisions;
•
Developing age-specific filters
, recognizing that different levels of religious
knowledge require tailored educational exposure.
In conclusion, this research demonstrates that AI
—
when guided by sound
religious ethics and cultural understanding
—
can serve not only as a digital shield
https://scientific-jl.com/luch/
Часть
-47
_ Том
-1_
июнь
-2025
436
against misinformation, but also as a
facilitator of responsible and meaningful
religious learning
in the online world.
References
1.
Devlin, J. et al. (2019).
BERT: Pre-training of Deep Bidirectional
Transformers for Language Understanding
. arXiv.
2.
Al-Darwish, S. (2021).
Religious Education and Youth in the Digital Age
.
Journal of Islamic Studies.
3.
Chowdhury, G.G. (2010).
Introduction to Modern Information Retrieval
.
Facet Publishing.
4.
Zannettou, S. et al. (2020).
Detecting and Characterizing Extremist Content
in Online Platforms
. ACM Transactions on the Web.
5.
Islamic Development Bank. (2023).
Ethics of AI in Islamic Education
.
6.
Vaswani, A. et al. (2017).
Attention Is All You Need
. NeurIPS.
7.
The Organisation of Islamic Cooperation (OIC). (2022).
Guidelines for Safe
Digital Religious Education
.
8.
Khan, S., & Rehman, A. (2020).
Filtering Hate Speech and Extremist
Content using AI
. IEEE Access.
