Intelligent Taxonomy Manager
Content Auto-tagging Manager
Knowledge Browser
Contact us
Software
Mondeca’s product offering includes ITM for taxonomy management, CAM for auto-tagging, and KB for knowledge publication. ITM, CAM, and KB are 100% web-based, browser-based products and run on an application server – absolutely no installation is required on your client machines. Our loosely-coupled software applications are designed to boost productivity and improve the way information is retrieved, analyzed and used. Web services expose key features and functionalities to interact with enterprise applications/systems in place.
Intelligent Taxonomy Manager
Intelligent Taxonomy Manager is a data-model driven, web-based taxonomy management application which adapts to any industry and can adjust to changes in data and knowledge in time. It allows the creation of an unlimited number of complex terms, properties and relationships – in an unlimited number of languages – which you are free to edit directly at any time.
Your Challenges
- Organize collaborative design and eliminate data silos
- Extend semantic standards to adjust to your business goals
- Define crosswalks within and across taxonomies
- Synchronize taxonomies with outside sources
- Manage suggestions for taxonomy enrichment
- Import and export data using open standards and formats
Design, maintain and publish taxonomies and ontologies adapted to your business, domain, language, or organization.
We help build, implement, maintain and integrate world-class, intuitive, and comprehensive navigation structures required to optimize product discovery when searching the client site.
Terminology server
ITM’s APIs provide access to multiple terminologies, classifications or code systems.
ITM key features
Manage multiple taxonomies
- Split data into workspaces
- Manage multiple taxonomies
- Set up access rights for each workspace
Multiple languages
- Any language and character set
- Alignment of multiple language taxonomies
Access rights
- Fine grained user profiles
- Integrates with enterprise IAM systems
Import / export data
- CSV, Excel
- Semantic standards: OWL, RDF, XML, SKOS, XTM, RDF/XML, N-Triples, Turtle, N3, JSON-LD, and RDF/JSON
Browse
- Easy navigation
- Tree or graph based data representation
Query and search
- Advanced search UI
- Multicriteria search
- Saved search
- Search based imports & exports
Improve data
- Candidate & suggested terms management
- Candidate term API
- Maintenance tasks assignement
Alert & track
- Event based alerts
- Event based data push
- Audit trail
- Activity dashboards
Open access to your data
Stop losing time building and searching for reference data. Establish a single source of truth for your teams, partners, clients and systems to get rid of inconsistencies. You can start out simple and gradually ramp up with more complex data representations to take up new challenges in time.
Key components
Intelligent Taxonomy Manager is based on a 3-tier architecture which dynamically adapts its UIs and APIs based on the business logic defined in the data model.
Your data is invaluable
Whether you are maintaining a simple flat list of terms or a full-blown ontology, you need tooling to support effective data management, regardless of data formats and languages, regardless of volumes and complexity of the data model. With ITM you can ramp up quickly and also stay the distance when things get harder. You can tailor your data so that it specifically fits your domain or industry. And down the line ITM will help you multiply the value of your data by placing it at the heart of your enterprise information system.
Intelligent Taxonomy Manager’s REST APIs support tight integration with third-party applications. It includes native connectors to CAM and KB and SharePoint Online TermStore
Query
Semantic network query
Manage
Semantic Network Management
Model
Data model access
Workspace
Workspaces & Users Management
Candidate
Candidate terms management
Log
Access ITM event log
Configure
Create and configure alerts
Content Auto-tagging Manager
Content Auto-tagging Manager is a an open, highly configurable and parameterizable semantic tagging pipeline based on a variety of processing engines. It takes advantage of your taxonomies and enables NLP/NLU analysis, relevancy scoring, SPARQL rule-based classification and machine learning into a single platform
Your Challenges
- Implement tagging processes at the onset of the content value chain
- Support manual indexing and/or override of suggested tags
- Bulk extract meaningful terms to create new taxonomies
- Connect to a variety of content types an sources, in multiple languages
- Enable image/video/voice metadata generation technologies
- Integrate with search and document management
- Minimize integration costs
- Automatically scale up to cope with a growing throughput of content
CAM includes optional and extensible connectors to websites, social media content sources and cloud-based APIs for image/video recognition and speech to text processing. It can flexibly process various media assets and trigger specific processing plugins.
Language Support
Additional languages may be quickly added at customer request. Language support means full linguistic support based on the language specific grammar. CAM can leverage term sets or dictionaries in any language and alphabet.
Integrate semantic tagging in your core business applications and processes. Dynamically enrich your taxonomies by extracting meaningful data from content.
Tagging multilingual content
CAM detects entities and relations in more languages than Google, Microsoft, Amazon and IBM together.
Text is a rich source of information, but categorizing it takes time and efforts due to its unstructured nature. We help implement automated categorization processes. We helped categorize news for media organizations, incoming documents for government organizations, consumer feedback for companies and many others.
Machine learning with CAM
Unleash the power of the latest deep learning models
Solve your NLP tasks with state of the art Transformer models like BERT
When to use
When the rules required to properly execute the task are not known or to complex to formulate. You will also need an annotated set of data with at least a hundred examples for each category.
What can you do with machine learning models
Classify sentences or documents, identify named entities or concepts, detect sentiment and others.
How
Create and train your own model. Alternatively, use the transfer learning approach – take an existing complex like a Transformer model and retrain it for your task.
Use Jupyter notebook to define and train Python ML models. Here is an example of notebook for a BERT based Tensorflow model.
Train deep learning models on environments like Google Colab. Remember, training will work best on a GPU fitted server.
Once your model is ready, use the CAM AdminUI to import it and set up the serving environment.
Supported models
Gate plugins
- Naive Bayes, Maximum Entropy, and Decision Trees (MALLET)
- linear chain Conditional Random Fields (MALLET)
- Support Vector Machines (LibSVM)
Python models including neuronal deep learning models
- scikit-learn
- spaCy
- Tensorflow
- PyTorch
- Transformer models: HuggingFace
Enjoy the machine learning in an industrial grade production setup
How to use
01
Select the algorithm
Get annotated data set
Split into training and validation
Train using the training data set
Evaluate results using validation set
Deploy to production
CAM key features
Persistence
- Store your work in the application persistence layer to support iterations and fine tuning of manual indexing work
Configuration
- Use existing templates for the creation of new resources (profiles, connectors, scripts, workflows, and engines)
- Edit and reload CAM central configuration without server restart
- View classification rules, RDF resources, and gazetteers
Integration
- Use CAM REST web service to execute auto-tagging workflows from content-centric applications
- Connect to CMS: Sharepoint, Drupal, WordPress
- Search engines: Elastic, SolR
- Graph databases
Workbench
- Execute, control and review results with informative and visual information
- Modular display: select side by side panels or widgets when reviewing results
- Monitor and configure CAM directly from the administration UIs
Execute multiple analyses
- Analyze content using NLP powered by business taxonomies
- Classify content using SPARQL classifications rules
- Execute Machine-learning to train CAM on a corpus of documents – and apply ML at sentence, paragraph or document level
Learn from content
- Analyze a corpus of several documents in one go
- Automatically suggest candidate terms to ITM for taxonomy improvement
- Bulk extract terminology from a corpus with TF/IDF scores
- Run difference comparison over 2 analyses, review the differences and compute precision/recall metrics either at a document level or at a corpus level
Application Monitoring
- Track workload and activities from the user and administrator dashboards
- Server nodes activity and storage capacity details
Permissions & Security
- Editable and granular rights for profiles
- Optionally map permissions to Mondeca Identity and Access Management component
Speed up indexing work and free up quality time for assessment
Enable automation of mundane and time-consuming indexing processes while allowing manual, qualitative review of machine-processed content
Key components
Content Auto-tagging Manager is based on a 3-tier architecture and can be either fully automated using the CAM API from another system or involve human agents using the generic web-based Auto-tagging UI.
Content Auto-tagging REST APIs support tight integration with third-party applications. It includes native connectors to ITM, KB, Solr, Elastic Search and SharePoint Online.
Process
Process a document through CAM workflow
List
List available workflows
Open/Close
Open or close a workflow
Configure
Set upCAM workflow parameters
Configure
Set up CAM workflow parameters
Already using CAM ?
Ugprade for V3 to get
- Enhanced connectivity
- Persisted indexing results
- Machine learning
- And more ….
Knowledge Browser
Knowledge Browser is a web-based portal featuring powerful search capabilities combined with graphical visualizations. It provides intuitive, read-only access to broad audiences who need to visualize, search, navigate and browse collections of enterprise data.
Your Challenges
- Publish reference data using open standards and formats
- Enable internal/external access to enterprise data via any web browser
- Organize and control the type and amount of data shared with clients
- Capture client requests for terminology improvement
- Aggregate data sources to streamline integration with applications
Browse, search & find, explore & discover, share & publish your data
KB key features
Browse
- User-friendly ultra-intuitive access to data.
- Navigate flat, hierarchical or graph-based terminologies
Explore and discover
- Configurable home dashboard for quick access to predefined datasets
- Graph-based visualization
Search
- Search enhanced through smart options
- Auto suggestion and query interpretation
- Multicriteria search
Find
- Refine search results based on facets (properties, types and data sources)
- Sort results according to relevancy
Share & Publish
- Configurable web portal
- Includes connectors to ITM and CAM
- Based on standards REST services
- Users can provide feedback, ask questions and download data
Move away from spreadsheet exchanges and think Graph
Let systems and targeted audiences take advantage of your enterprise data using semantic standards for data publishing and dissemination
Key components
Knowledge Browser is based on a 3-tier architecture which combines the powers of a graph database and a search engine index in the backend.
Knowledge Browser’s REST APIs support tight integration with third party applications. It includes native connectors to ITM and CAM.
Publish
Send data to Knowledge Browser
Fetch
Grab data from Knowledge Browser
Configure
Update configuration of Knowledge Browser
Building blocks
Connectors
Inbound
CAM features a large array of connectors to content sources: web scraping software, social media APIs, Content Management Systems such as SharePoint as well as enterprise databases, Digital Assets Management tools, file systems and web-based storage.
Outbound
Both ITM and CAM have significant outbound connectivity. CAM results can be published in real time to search engines, business applications and Content Management Systems. The ITM SharePoint TermStore connector lets you seamlessly synchronize your enterprise taxonomies with you SharePoint environment.
3rd party services
CAM connects to third party services to execute specific workflow steps or enrich information. This includes visual recognition and voice analysis services like Clarifai, IBM Watson or Speechmatics. Concepts and terms managed by ITM can be enriched through linked data sources.
CAM Auto-Tag SharePoint add-in
Trigger auto-tagging on demand or when a document is added to a documents library. The CAM add-in sends the SharePoint document to CAM for auto-tagging, gets tags from CAM and adds them in different documents columns (e.g. ‘tags’ or ‘classification’) based on the add-in configuration.
The add-in is implemented as a Microsoft Power Automate flow for Microsoft SharePoint Online. Other Microsoft SharePoint versions can be supported if needed.
Semantic Storage
Graph Database
RDF
ITM and CAM are RDF compliant applications, which opens to a wealth of potential uses, for both internal data management and interoperability with the Web of Data at large.
Neo4j
Knowledge Browser information is stored as Neo4j graphs
Search Engine
Elastic
Information can be published to Elasticsearch. Elasticsearch engine is also integrated into Knowledge Browser architecture.
SolR
Information can be published to Solr.
Access Control
Identity and Access
Mondeca uses KeyCloak for the identity and access management service for applications. Mondeca supports Single-Sign On and Single-Sign Out for browser applications and synchronization of users from LDAP and Active Directory servers.
Delivery mode
Software as a Service
Subscribe to Mondeca software in the cloud. We take care of the infrastructure, software installation, upgrades, operational support and capacity scaling. Your environment is secure, either on AWS or OVH, depending on your preference. It is your own environment – our policy is that nothing is shared between clients.
In your own private cloud
Integrate our software into your own private cloud. We will manage installations, configuration, upgrades, scaling and operational support within you private cloud environment. Your data will stay within your security perimeter and you will keep the control of your cloud set up.
On premises
Perpetual software licence and annual maintenance and support contract. We will train and assist your teams with installation and configuration tasks.
High security
Please get in touch with us if you require a high security environment, including air gap.
Cloud-based architecture of Mondeca’s solution
Example of a Mondeca setup using AWS Virtual Private Cloud
The proposed AWS architecture is highly available and secured. This architecture can be simplified or completed based on specific needs (high-availability, increased security with VPN access). This architecture and the choice of the AWS deployment region and Availability Zones is always adapted to our clients requirements.
Integration
National Public Radio
This video shows how Mondeca supports auto tagging for NPR Research, Archives and Data Strategy team (RAD). Tags and relevancy scores are generated behind the scenes and on the fly by CAM, received seamlessly by NPR RAD Artemis content management system and quickly presented for user review. The consistency of tagging is controlled by ITM. This is an example of automated tagging validated by expert users.