A Digital Solution Bangladesh Initiative
A dedicated programme for building open, standards-based digital libraries, institutional repositories and research archives — so that the knowledge held by Bangladesh's universities, ministries, courts and cultural institutions is discoverable today and readable decades from now.
A digital library is not a website with PDFs on it. It is a catalogue, a preservation plan and a set of open standards working together — and that is exactly what we build.
The Digital Library Initiative grew out of the library automation work Digital Solution Bangladesh has delivered for the Prime Minister's Office, the National Book Center, Jagannath University and the Chapainawabganj District Judge Court. Every one of those projects raised the same questions — how should a Bangla-language collection be catalogued, how do you retro-convert decades of card records, and who can still read the files in 2045?
Rather than answer those questions once per client, we made them a standing programme. The initiative pairs production engineering with applied research and an academic advisory team, so that what we learn on one deployment becomes reusable practice for the next institution — documented, open and free of vendor lock-in.
Two commitments guide every decision in this programme — from which metadata schema we pick to which file format we archive in.
To make the recorded knowledge of Bangladesh's institutions openly discoverable, properly catalogued and safely preserved — by deploying open-source library platforms, applying international metadata standards, and equipping library staff to run these systems confidently on their own.
A connected national knowledge network in which every university, ministry, court and research institute in Bangladesh runs an interoperable digital library — where a single search reaches across collections, Bangla scholarship is as machine-readable as English, and no institution's archive is trapped inside proprietary software.
Open-source platforms, open metadata standards and open protocols — the institution owns the system, the data and the exit route.
We agree the cataloguing rules and metadata profile first. The software is chosen to serve that decision, never the reverse.
Professional librarians define policy and workflow; engineers implement it. Advisory oversight keeps that balance honest.
Checksums, archival masters, documented formats and a migration plan — storage alone is not preservation.
Eight service lines that together take an institution from a card catalogue and a store room to a searchable, citable, preserved digital collection.
Koha deployment covering acquisitions, cataloguing, circulation, serials, reservations, fines and branch management — configured to your service rules, not to a demo default.
DSpace repositories for theses, dissertations, journal articles, reports and datasets — with communities, collections, embargo rules and persistent handles for citation.
Scanning workflows for rare books, manuscripts, gazettes and judgments, producing archival masters plus searchable text through Bangla and English OCR pipelines.
Original and copy cataloguing in MARC21 and Dublin Core, authority control, subject headings, and conversion of legacy card or spreadsheet records into clean bibliographic data.
A fast, mobile-first OPAC with faceted search, Bangla transliteration handling, saved lists and a reading interface that works on a phone over a slow connection.
OAI-PMH endpoints, Z39.50 and SRU targets, SIP2 for self-service kiosks, and single sign-on against your existing identity system so the library is not an island.
Format policy, fixity checking, versioned archival storage, off-site replication and a documented migration path — designed against OAIS principles.
Hands-on sessions for cataloguers, circulation desk staff and system administrators, with written SOPs in Bangla and English so knowledge stays after the project closes.
The initiative is guided by academics and practising librarians. They review our technical direction, challenge our assumptions and keep the programme anchored in library science rather than in software fashion.
Advises the initiative on its applied-computing direction — machine learning for discovery and recommendation, OCR and document-understanding pipelines, and the responsible use of automation in cataloguing. Also guides how the programme's research questions are framed and how results are evaluated.
Brings the professional library perspective — collection development, cataloguing and classification practice, reader services, and institutional repository policy. Reviews our metadata profiles and workflow designs against how academic libraries in Bangladesh actually operate day to day.
We are expanding the board with specialists in archival science, digital preservation and open-access publishing. If you work in one of these areas and would like to contribute to a national digital library effort, we would welcome a conversation.
Express interestAdvisors meet with the engineering team at each programme milestone. They review metadata profiles and preservation policy before a system goes live, sign off on the research agenda for the coming cycle, and give an independent read on whether a proposed deployment genuinely serves the institution's readers. Their guidance is advisory and non-binding, but we publish how it shaped each decision.
Six active research tracks. Each one exists because a real deployment ran into a problem that the off-the-shelf answer did not solve well enough for Bangla-language collections.
Recognition accuracy for Bangla drops sharply on aged paper, conjunct-heavy typesetting and typewritten government records. We are benchmarking pre-processing and recognition pipelines against real scanned material and publishing what actually works.
Can a model draft a MARC21 record from a title page and leave a cataloguer to verify it? We are measuring where suggestion saves time, where it introduces error, and which fields should never be automated.
Readers search in Bangla, in English and in romanised Bangla — often in one query. We are testing transliteration handling, stemming and embedding-based retrieval so that all three reach the same record.
OAIS-aligned preservation is usually costed for well-funded archives. We are documenting a realistic tier for Bangladeshi institutions: which formats, which checksum cadence, which replication strategy, at what budget.
Harvesting several repositories into one search sounds simple until the metadata disagrees. We are working on a shared Dublin Core application profile and de-duplication rules for a future national union catalogue.
Libraries need circulation and access statistics; readers deserve privacy. We are designing aggregate-only analytics that inform collection decisions without building a per-reader behaviour record.
We work with departments, library schools and student researchers on these tracks — through supervised projects, shared datasets and joint pilots on live collections. Findings, test corpora and configuration recipes are released back to the community wherever licensing and institutional consent allow.
Nothing here is proprietary. Every standard below is published, widely implemented and supported by more than one piece of software — which is what makes an exit possible.
Every system we deliver can export its full bibliographic and item data in a standard format on demand. Source code, configuration and documentation are handed over at project close. If you ever choose to leave us, nothing about the platform makes that difficult — and we consider that a feature, not a risk.
Five phases. Each one ends with something the institution can inspect and sign off — never a long silence followed by a launch.
We survey the collection size and condition, existing records, staffing, network and space. You receive a written assessment with a realistic scope, a metadata policy proposal and an honest view of what should not be digitised.
Koha and DSpace are installed, hardened and configured to your circulation rules and community structure. The cataloguing framework, controlled vocabularies and Dublin Core profile are agreed with your librarians and the advisory team before any records are loaded.
Scanning runs in batches with quality control at each one. Legacy card, register and spreadsheet records are converted to MARC21, de-duplicated and checked against authority files. Bangla and English OCR produce the searchable text layer.
The public catalogue is themed and tested on real devices, OAI-PMH and Z39.50 endpoints are opened, single sign-on and self-service hardware are connected, and the system goes live alongside a fallback until staff are comfortable.
Role-based training for cataloguers, desk staff and administrators, with bilingual SOPs. Then scheduled backups, fixity verification, version upgrades and priority support under a maintenance agreement.
The platform is the same; the cataloguing policy, access rules and preservation priorities are not. Each of these needs a different answer.
Course reserves, thesis and dissertation repositories, faculty publication records, and open-access policy support for departmental research output.
Policy documents, gazettes, project reports and circulars — organised for officials who need the right version of a document quickly and with confidence.
Judgments, law reports, statutes and case files with citation-aware search and strict access control over what is public and what is not.
Rare books, manuscripts, periodicals and photographs, handled with preservation-grade capture and description suited to unique, fragile material.
What librarians and administrators ask us most often.
Not always. Koha manages the physical and licensed collection — acquisitions, circulation, serials. DSpace holds digital objects your institution produces, such as theses, reports and datasets. A library with no born-digital output may only need Koha; a research centre with no lending desk may only need DSpace. We recommend after the assessment, not before it.
Cataloguing, display and search work in Bangla throughout, and the interface can be delivered bilingually. OCR accuracy on Bangla varies with the source: clean modern print performs well, while aged paper, dense conjuncts and typewritten records are harder — which is exactly why Bangla OCR is one of our active research tracks. We report expected accuracy on a sample of your own material before committing to a digitisation plan.
They are migrated, not retyped. Card catalogues, registers, spreadsheets and older software exports are converted into MARC21, de-duplicated and enriched through copy cataloguing from external Z39.50 targets. You approve a sample batch before the full conversion runs, and the original files are retained untouched.
Yes. On-premise deployment inside your own data centre is fully supported, and for many government and judicial clients it is the requirement. We can also host, or run a hybrid where the catalogue is on-premise and preservation copies are replicated off-site. Backup and fixity procedures are documented either way.
The institution does — content, metadata, configuration and source code alike. We work on open-source foundations and open standards specifically so that ownership is real rather than contractual. Full exports in standard formats are available at any time, not just at the end of an engagement.
Circulation history retention is configurable and defaults to the minimum the library needs. Analytics are aggregate by design — we do not build per-reader behaviour profiles. Access to borrower records is role-restricted and audited, and the policy is written down and handed over rather than left to system defaults.
Yes — that is a stated aim of the initiative. We host supervised student projects, share anonymised test corpora and run joint pilots with departments and library schools. Get in touch with a short outline of what you would like to investigate and we will route it to the relevant advisor.
Tell us what you hold and who needs to reach it. We will come back with an assessment approach, a realistic scope and an honest view of the effort involved — no obligation.
