Hadith | Dr Riasat Islam

Designing Islamic Apps for Interconnected Learning

Thu, 16 Jul 2026 14:40:00 +0100

Our preprint, Designing for Interconnected Islamic Learning: A Qualitative Study of Muslim Women’s Experiences with Qur’an, Hadith, and Seerah Apps, is now available online.

The paper is currently available as an arXiv preprint and is under review with a reputed journal. That means the work is public and readable, but it should still be treated as a preprint until peer review is complete.

You can read the publication page here: .

Why interconnected Islamic learning matters

Islamic learning is rarely isolated.

When people read the Qur’an, they often want to understand context, language, tafsir, related hadith, Seerah background, and practical lessons. When they read a hadith, they may want to know where it sits within wider Islamic teaching. When they study the Seerah, they may want to connect events to revelation, character, worship, and everyday life.

But many digital tools separate these learning pathways.

One app may focus on Qur’an reading. Another may focus on Hadith. Another may provide Seerah material. Each app may be useful, but the learner has to do the work of connecting them. They move between screens, search boxes, translations, bookmarks, and notes, often without a clear sense of how the pieces fit together.

Our study looks at this problem from a human-computer interaction perspective. It asks: how do Muslim users actually move between Qur’an, Hadith, and Seerah apps, and what kinds of design support would make Islamic learning more connected without making it overwhelming?

What the study explored

The paper is based on five semi-structured interviews with Muslim women who use Islamic apps for learning and reflection.

The aim was not to measure app popularity or compare feature lists. The aim was to understand lived experience: how people search, read, cross-check, reflect, and decide whether an Islamic learning tool feels useful and trustworthy.

Several recurring tensions appeared in the interviews:

users want more context, but not clutter
users value guidance, but do not want the app to overstep
users want connected sources, but still need clear provenance
users want flexibility across quick reading and deeper study
users want authenticity without losing devotional flow

These tensions matter because Islamic learning apps are not just information products. They shape how people encounter sacred knowledge, build habits, ask questions, and develop confidence.

Layered contextuality

One of the key ideas in the paper is layered contextuality.

By layered contextuality, we mean that Islamic learning tools should provide context in layers rather than all at once.

A user reading a Qur’anic verse may first want a simple translation. Then they may want a short explanation. Then they may want related hadith. Then they may want Seerah background, scholarly commentary, or practical application. Another user may only want the first layer at that moment.

The design challenge is to make this expansion trustworthy, optional, and respectful of the user’s purpose.

Good contextual design should not flood the screen. It should not turn every moment of reading into a maze of links. It should help users move deeper when they are ready, while preserving clarity for those who want a focused devotional experience.

This is especially important on mobile devices, where attention, screen space, and time are limited.

Trust is a design requirement

In Islamic apps, trust is not a nice extra. It is a core requirement.

Users need to know where content comes from. They need to distinguish between Qur’anic text, translation, hadith, commentary, summary, and app-generated guidance. They need confidence that the app is not mixing sources carelessly or presenting interpretation as if it were direct revelation.

This becomes even more important as Islamic apps begin to use AI, semantic search, recommendation systems, and generated summaries.

The more powerful the tool becomes, the more important provenance becomes.

A useful Islamic learning app should answer questions such as:

What source is this content from?
Is this a translation, commentary, summary, or generated explanation?
Can I inspect the original reference?
Is the app helping me understand, or pushing me into more content?
Does the design preserve adab with sacred knowledge?

These are not only theological questions. They are design questions.

Devotional flow and study modes

Another important theme is that users approach Islamic learning in different modes.

Sometimes a person wants a quick reminder. Sometimes they want to read after prayer. Sometimes they want to study seriously, compare sources, take notes, or prepare for teaching. A single app may need to support more than one mode without making the experience confusing.

This is where devotional flow matters.

If every verse, hadith, or Seerah event is surrounded by too many links, popups, prompts, and recommendations, the app may technically provide more information but spiritually feel less helpful. More content does not always mean better learning.

The design goal should be controlled depth.

Users should be able to stay focused when they want focus, and expand context when they want depth.

What Islamic app builders can learn

For developers, designers, researchers, and Islamic institutions, the paper points toward a few practical principles.

First, design around learning journeys, not only content categories. Qur’an, Hadith, and Seerah are connected in real learning, even if they are stored separately in a database.

Second, make context progressive. Start simple, then let users expand into deeper layers.

Third, make sources visible. Trust increases when users can see where content comes from and what kind of content it is.

Fourth, avoid interface clutter. Islamic learning apps should not copy every engagement pattern from social media or productivity apps.

Fifth, support different intensities of learning. Quick reflection, daily reading, structured study, and deeper research are not the same use case.

Sixth, involve real users. Good Islamic technology cannot be built only from assumptions about what Muslims need. It requires listening carefully to how people actually learn, struggle, search, and reflect.

Why this connects to Islamic Computing

This paper fits into a wider direction I have been calling Islamic Computing: the use of computing, design, data, AI, and research to build tools that support Muslims in learning, worship, character, and service.

Islamic Computing is not only about digitising Islamic content. It is about asking how technology should be shaped by the nature of the knowledge it carries.

If the knowledge is sacred, the design must be careful.

If users are learning across Qur’an, Hadith, and Seerah, the system should support connection without confusion.

If people rely on the app for religious understanding, then trust, provenance, humility, and scholarly seriousness need to be built into the product from the beginning.

In summary

The preprint argues that Islamic learning tools should support connected movement across Qur’an, Hadith, and Seerah while preserving trust, optionality, and devotional flow.

The future of Islamic apps should not only be more content or more features. It should be better learning architecture: clearer sources, layered context, respectful interaction, and design that helps Muslims understand and practise Islam with confidence.

Read the preprint details here: .

The arXiv record is also available directly at .

Hadith Computational Science in the Age of Large Language Models

Thu, 16 Jul 2026 14:30:00 +0100

Our new preprint, Hadith computational science in the age of large language models: a critical narrative review, is now available online.

The paper is co-authored with Md. Ashraful Haque and is currently available as a preprint on Zenodo. It is also under review with a reputed journal. That means the work is public and citable, but it should still be read as a preprint until the journal review process is complete.

You can read the publication page here: .

Why this paper matters

Hadith collections are among the most important sources of Islamic knowledge. They are also complex textual ecosystems. A single hadith may involve Arabic wording, translation, narrator chains, grading, commentary, legal interpretation, historical context, and relationships with other narrations.

This makes hadith a serious challenge for computational systems.

Search engines, databases, embeddings, and large language models can help people find and explore hadith material. But they can also create risk if they remove context, blur provenance, generate unsupported claims, or present uncertain outputs with false confidence.

The central argument of the paper is that progress in hadith computational science should not be measured only by benchmark scores or impressive demonstrations. It should also be measured by the quality of the evidence infrastructure around the system.

In practical terms, that means asking:

Where did this answer come from?
Which text, translation, edition, or dataset was used?
Can the result be reproduced?
Can a scholar or expert inspect the reasoning path?
Does the system communicate uncertainty clearly?
Is it supporting Islamic scholarship, or replacing careful scholarship with a thin technical shortcut?

These questions become more urgent in the age of large language models.

From keyword search to retrieval-grounded AI

Older hadith search systems often depended on keywords, metadata, or structured databases. These systems are useful, but they can struggle when people search by meaning rather than exact wording.

Large language models and transformer-based systems change the interface. Users can now ask conversational questions, request summaries, compare narrations, and explore themes across collections. Retrieval-augmented generation can also connect a generated answer to specific source passages.

This creates a real opportunity. A well-designed system could help students, researchers, developers, and institutions navigate hadith sources more effectively.

But the opportunity comes with a responsibility.

Hadith is not ordinary text. It is not enough for a model to produce a fluent answer. In Islamic knowledge work, fluency without traceability is dangerous. A system must be able to show what it used, where it came from, and how much confidence users should place in the output.

This is why the paper gives attention to provenance, reproducibility, expert supervision, and evaluation design.

What we mean by hadith computational science

By hadith computational science, we mean the use of computational methods to collect, structure, search, analyse, compare, retrieve, and present hadith-related material.

This can include:

digitising and structuring hadith collections
modelling narrator networks and isnad data
linking hadith to commentary, fiqh, Seerah, and Qur’anic themes
evaluating hadith search engines
building multilingual retrieval systems
using natural language processing for classification or semantic search
designing AI systems that support Islamic learning and scholarship

The field sits between computer science, Islamic studies, digital humanities, information retrieval, natural language processing, and human-computer interaction.

That interdisciplinary nature is important. A purely technical approach can miss scholarly nuance. A purely manual approach can struggle with scale, multilingual access, and modern search expectations. The best future work will need both technical competence and serious Islamic scholarly input.

The main concern: trust

Many conversations about AI focus on capability. Can the model answer the question? Can it summarise a text? Can it retrieve the right passage?

For hadith systems, capability is only one part of the problem.

The deeper issue is trust.

A trustworthy hadith AI system should not behave like a black box oracle. It should behave more like a careful research assistant: useful, limited, transparent, and reviewable.

That means it should cite sources. It should preserve context. It should distinguish between text, translation, commentary, grading, and interpretation. It should make it clear when an answer is generated, when it is retrieved, and when expert review is needed.

This also means that evaluation should go beyond whether a system returns a plausible-looking answer. We need evaluation methods that test retrieval quality, source fidelity, hallucination risk, translation reliability, and the ability to support expert review.

Why scholar-guided evaluation is essential

Hadith work has established scholarly disciplines, methods, and standards. Computational systems should not ignore that.

A model may identify linguistic similarity between two narrations, but that does not mean it has understood legal relevance. It may retrieve a hadith that looks topically related, but that does not mean it has selected the strongest evidence for a ruling. It may summarise a narration fluently, but still flatten important context.

This is why scholar-guided evaluation is essential.

The goal is not to slow down useful technical work. The goal is to make the work reliable enough to be genuinely useful.

For me, this connects to a wider theme I have been thinking about under the term Islamic Computing: how Muslims can build technologies that are not merely impressive, but beneficial, trustworthy, and aligned with the responsibilities of the knowledge they handle.

What developers and researchers can take from the paper

If you are building systems for Islamic knowledge, this paper suggests a few practical principles.

First, design for provenance from the beginning. Do not treat citation and source tracking as features to add later.

Second, separate retrieval from generation. Users should know which parts of an answer came from source documents and which parts were generated by the system.

Third, make outputs inspectable. A scholar, researcher, or advanced user should be able to trace the evidence path.

Fourth, evaluate failure modes. In this domain, a system that is usually right but confidently wrong in sensitive cases can still be harmful.

Fifth, build with humility. AI can support Islamic learning and research, but it should not be positioned as a replacement for scholarship, adab, or careful human judgement.

In summary

The preprint argues that large language models can reshape hadith computational science, but only if the field develops stronger standards for evidence infrastructure, provenance, reproducibility, and expert supervision.

The future of Islamic AI should not be driven by novelty alone. It should be driven by trust, transparency, scholarly seriousness, and benefit.

Read the preprint details and DOI here: .

The Zenodo record is also available directly at .