Technical roadmap

DESTINY is developing a technical roadmap to provide a structured overview of software development activities within the scope of DESTINY and associated projects. The roadmap will show the current state of software capabilities, planned developments, and longer-term technical ambitions for our open data infrastructure and Digital Evidence Synthesis Tools (DESTs).

The roadmap is intended to support transparency, coordination, strategic planning and community engagement around the development of open and sustainable software infrastructure for evidence synthesis.

The roadmap focuses on software and technical capabilities developed through DESTINY and recognises important contributions made by associated projects. It reflects the collective impact ambition of the Evidence Synthesis Infrastructure Collaborative (ESIC) and recognises the dependencies of other community projects on the technology which DESTINY is developing.

The roadmap will be reviewed and updated quarterly to reflect completed developments, changing priorities, emerging opportunities and lessons learned through implementation. The first edition of the roadmap will be published here in Q4 2026. For now, we are happy to share an indication of the tools and features currently in scope and under development.

Living evidence repository

Abstract retrieval

Coming soon

Bulk importer

A way to import records in bulk from OpenAlex snapshots

Climate & Health classification pipeline

An automated pipeline which searches for, assesses relevance of, and tags with taxonomy terms climate & health literature

Climate & Health relevance classifier

A classifier which predicts whether a title & abstract are relevant to the domain of Climate & Health

Climate & Health taxonomy tagger

A process which will predict tags from the DESTINY taxonomy which describe a title and abstract

Closed-access abstract mining

A way to retrieve and mine closed-access abstracts for inclusion within repository search, classification, and evaluation

Community administration UI

An administration interface which allows the configuration of the component parts of a community and its repositories (annotations, automations, taxonomies)

Core ontology

A developed core ontology which supports the basic elements of evidence synthesis

CSV importer

Bespoke importers which allow bibliographic, taxonomic, and extraction data to be imported to the repository as “raw” enhancements

Deduplication toolkit

A highly accurate, tested deduplication algorithm packaged as a python library for integration into robots and the repository

Demonstrator UI

Prototype user interface over the repository data to demonstrate the capabilities and test user interface patterns. Deprecated in 2025 in favour of the user interface components built more recently.

Faceted Search

Allows the dynamic filtering of search results using taxonomy tags

Granular permissions

Access control which limits a user’s ability to interact with resources from other communities

Import API

Import of data from external sources (including OpenAlex, extracted data, bibliographic data)

Incremental importer

Daily updates to the repository based on new records in OpenAlex

Integrated deduplication

Records are automatically deduplicated across the entire repository on import. Deduplication is non-destructive.

Python SDK

A software development kit which makes it easier to build against the repository infrastructure

Relevance classifier

Classifier which determines whether a reference is relevant to the “Climate & Health” domain based on title and abstract

Relevance robot

A robot which integrates the relevance classifier with the repository using the robots protocol

User-driven full text uploading

A way for users to upload relevant full texts not fetched by automated processes for analysis

Data extraction evaluation toolkit

Automated taxonomy extraction and evaluation

DEET, the data extraction evaluation toolkit, can use taxonomy/vocabulary hierarchical structure to optimise prompts for large taxonomy extraction.

DEET UI

DEET, the data extraction evaluation toolkit will have a web UI that allows users to build and evaluate automated data extraction pipelines interactively.

Evaluation of classification with LLMs

DEET, the Data Extraction Evaluation Toolkit can be used to classify documents using LLMs, and to evaluate how well LLMs can classify documents. Classification is useful for screening, or coding study themes.

Evaluation of data extraction for evidence synthesis with LLMs

DEET, the Data Extraction Evaluation Toolkit can be used to extract attributes of interest from published research documents using LLMs, and to evaluate how well LLMs can extract data. Data extraction is required for analysis of evidence.

Evaluation of armified data extraction using LLMs

DEET, the data extraction evaluation toolkit, will provide users with the means to test, develop, and evaluate automated approaches to armified data extraction from intervention studies.

Enhancements

EPPI-Reviewer importer

Coming soon

Full text retrieval

A robot which fetches full texts and abstracts from additional databases to supplement those gathered from OpenAlex

Generic robots

The ability to deploy a single robot whose behaviour can be determined by parameters supplied in the enhancement request

Geolocation robot

A robot which annotates references with mentioned geographic entities using the mordecai3 model.

Import search results

The ability to import results from a search in another database as bibliographic metadata into the repository, marking these as in-scope for a particular community.

Integrated full text fetching

A process which will fetch as many full texts as are available given publisher constraints given a user’s credentials

References API

Fetch enhancements associated with references

Robots API

Integration of enhancing processes to classify/enrich references in the repository

Scaling enhancement storage

Ensure that the storage of enhancements can scale effectively across multiple communities without limiting the ability for communities to experiment.

Streamlined EPPI imports

Easy import of data from EPPI reviewer which is aligned with a supported taxonomy

Transferability model

Model for determining whether a reference is relevant to policy-makers in Cameroon

Transferability robot

A robot which implements the transferability model as part of the repository ecosystem

User interfaces

Agentic mapping backend

Backend functionality that supports: a) agentic decomposition and transformation of a plain-language user query into a search (free-text, taxonomy, snowballing); b) agentic execution of search(es), formulation of screening criteria, and screening of search results; c) mapping of search results against either agentically derived or taxonomy terminology

Agentic mapping interface

Natural language system & interface to allow users to navigate the mapping and faceted search interfaces

Agentic mapping interface

Natural language system & interface to allow users to navigate the mapping and faceted search interfaces

Keyword search interface

An interface over the repository which allows boolean search over titles and abstracts, with short summaries of results shown in a paginated interface

Mapping interface

A cross tabulation which shows the frequency of documents matching two sets of taxonomy concept tags (based upon their concept schemes)

Mapping interface

A machine interface which presents published vocabularies for use in UIs, repository storage, and robots

Reference display

A rich display of the enhanced data of a result, including taxonomy tagging and extracted data

Search API

Keyword search and annotation filtering of references within the repository (including segmenting by sub-repository/domain)

Vocabulary builder interface

Tools for building simple OWL ontologies and SKOS taxonomies to form a vocabulary which defines linked data within the repository. Also includes an MCP to allow LLM assisted vocabulary construction

Vocabulary reader interface

An interface to present SKOS taxonomies to users and solicit feedback on them