Microsoft Language Understanding (LUIS) - Detailed Review
Analytics Tools
Microsoft Language Understanding (LUIS) - Product Overview
Microsoft Language Understanding (LUIS)
Microsoft Language Understanding (LUIS) is a cloud-based conversational AI service that enables developers to build applications that can interpret and respond to user input in natural language. Here’s a brief overview of its primary function, target audience, and key features:
Primary Function
LUIS uses machine learning to allow applications to receive user input in natural language, extract the meaning, and pull out relevant, detailed information. This service helps in building conversational interfaces such as chatbots, voice assistants, and other interactive systems that can understand and respond to user queries effectively.
Target Audience
The primary target audience for LUIS includes software developers, especially those working on conversational AI applications. It is particularly useful for large enterprises that need to create custom language models specific to their application domains without requiring extensive machine learning expertise.
Key Features
Custom Machine Learning Models
Developers can build custom language models iteratively, improving them based on real user traffic using advanced machine learning techniques.
Prebuilt Domain Models
LUIS offers prebuilt domain models that include intents, entities, and utterances, allowing developers to get started quickly. These models can be combined with custom information to suit specific needs.
Intent and Entity Identification
LUIS identifies user intentions (intents) and extracts relevant details (entities) from user utterances. For example, an intent like “BookFlight” can trigger actions that require entities such as travel destination, date, and airline.
Active Learning
The service uses active learning to improve model performance over time. It provides real utterances that it is unsure of for developers to review, label, and use to retrain the model.
Integration with Other Services
LUIS can be easily integrated with other Microsoft services like the Microsoft Bot Framework, QnA Maker, and Speech service, making it versatile for various conversational AI scenarios.
Security and Compliance
Backed by Azure infrastructure, LUIS offers enterprise-grade security, privacy, and compliance, ensuring that user data is encrypted and can be deleted at any time.
Important Note
It is worth noting that LUIS is scheduled to be retired on October 1st, 2025, and new LUIS resources cannot be created after April 1st, 2023. Microsoft recommends migrating LUIS applications to conversational language understanding for continued support and multilingual capabilities.
Microsoft Language Understanding (LUIS) - User Interface and Experience
Integration and Ease of Use
LUIS is integrated seamlessly with other Azure services, which enhances its ease of use for developers. For instance, it can be used in conjunction with the Azure Bot Service to create conversational interfaces such as chatbots and voice assistants. This integration allows developers to leverage LUIS’s natural language processing capabilities without needing to manage multiple disparate services.
Developer Experience
The service is known for its ease of use, particularly for developers familiar with Azure services. Developers can create, train, and deploy LUIS models using tools like the LUIS.ai portal or through the Azure Bot Framework SDK. This process involves defining intents, entities, and utterances, which can be managed through a user-friendly interface.
User Interface Elements
While the specific UI elements of the LUIS portal are not detailed in the provided sources, it is clear that the service is built to be accessible. Developers can annotate, train, evaluate, and deploy models using the Language Studio, which is part of the Azure AI services. This studio likely includes intuitive tools for managing language models, such as visual interfaces for intent and entity creation, and analytics tools to evaluate model performance.
Data Management and Privacy
From a user experience perspective, LUIS also provides clear controls over data management. Users have the ability to view, export, or delete any user content, aligning with GDPR policies. Additionally, users can opt to turn on or off query logging, giving them control over how their data is handled.
Overall User Experience
The overall user experience of LUIS is centered around facilitating the creation of intelligent, conversational applications. By providing a straightforward process for building and deploying natural language models, LUIS makes it easier for developers to focus on the logic and functionality of their applications rather than the intricacies of natural language processing. The service’s interoperability with other Azure tools and its clear data management options contribute to a positive and efficient user experience.
In summary, while specific UI elements are not detailed, the integration with Azure services, ease of use for developers, and clear data management options all contribute to a user-friendly and efficient experience when working with Microsoft’s Language Understanding (LUIS).
Microsoft Language Understanding (LUIS) - Key Features and Functionality
Microsoft Language Understanding (LUIS)
Microsoft Language Understanding (LUIS) is a cloud-based conversational AI service that leverages custom machine-learning intelligence to interpret and process natural language text. Here are the key features and functionalities of LUIS:
Simplicity
LUIS is designed to be user-friendly, allowing you to build conversational AI applications without requiring in-house AI expertise or prior machine learning knowledge. You can start with quickstarts or use prebuilt domain apps to get started quickly.
Natural Language Processing (NLP)
LUIS applies machine-learning intelligence to predict the overall meaning of a user’s conversational text and extract relevant, detailed information. This includes identifying intents (the user’s goal) and entities (specific details within the text).
Integration
LUIS can be easily integrated with other Microsoft services such as the Microsoft Bot Framework, QnA Maker, and Speech service. This integration enables you to create comprehensive conversational solutions that can handle various user interactions.
Security, Privacy, and Compliance
LUIS is backed by Azure infrastructure, which provides enterprise-grade security, privacy, and compliance. Your data remains encrypted while in storage, and you have full control over your data, including the ability to delete it at any time.
Customization
You can build custom LUIS applications by creating, training, and publishing LUIS apps. This involves defining intents, entities, and utterances that your model will recognize. LUIS also supports the use of prebuilt domains to speed up the development process.
Multilingual Support
Although LUIS itself does not support multilingual capabilities out of the box, its successor, Conversational Language Understanding (CLU), allows you to train models in one language and use them in multiple languages without retraining.
Orchestration
LUIS can be part of an orchestration workflow that connects multiple language understanding projects, question answering knowledge bases, and other conversational components. This allows you to build multi-purpose bots that can handle a variety of conversational topics.
Deployment and Management
LUIS provides access through its custom portal, APIs, and SDK client libraries. You can manage and deploy your LUIS apps using these tools, ensuring that your conversational AI solutions are scalable and maintainable.
Retirement and Migration
It is important to note that LUIS will be retired on October 1, 2025, and new LUIS resources cannot be created after April 1, 2023. Microsoft recommends migrating existing LUIS applications to Conversational Language Understanding (CLU) to benefit from continued product support and enhanced features.
Conclusion
In summary, LUIS offers a straightforward way to integrate AI-driven natural language processing into your applications, with a focus on simplicity, security, and integration capabilities, although it is being phased out in favor of more advanced technologies like CLU.
Microsoft Language Understanding (LUIS) - Performance and Accuracy
Evaluating the Performance and Accuracy of Microsoft’s Language Understanding (LUIS)
Evaluating the performance and accuracy of Microsoft’s Language Understanding (LUIS) involves several key aspects, as well as acknowledging some of its limitations and areas for improvement.Performance Metrics
The performance of LUIS is primarily measured by its ability to predict intents and entities accurately from user utterances. This is done by comparing the system’s predictions against a human judge’s evaluations. The metrics include true positives, true negatives, false positives, and false negatives. These metrics help developers assess how well the system recognizes custom natural language processing (NLP) concepts.Confidence Scores
LUIS uses confidence scores to indicate how certain the system is about its predictions. These scores range from 0 to 1, with higher scores indicating greater confidence. For instance, a score of 0.99 is highly confident, while a score of 0.01 is very low. The confidence score is directly influenced by the quality and variety of the training data provided during the application’s authoring phase.Improving Accuracy
To enhance the accuracy of LUIS, several strategies can be employed:Active Learning
Enabling the active learning feature allows the system to log user queries and select those that need validation. This helps in identifying and correcting incorrectly predicted utterances, which can then be added to the correct intent for retraining and republishing the application.Patterns
Using patterns, which are template utterances assigned to intents, can increase confidence scores for intent and entity predictions without requiring a large number of additional utterances.Balanced Training Data
Ensuring that the training dataset is balanced is crucial. Imbalance in the utterance training set can cause the LUIS model to predict the wrong intent. Balancing the dataset can improve precision and overall prediction accuracy.Regular Review
Periodically reviewing endpoint utterances and updating the model with new examples can help maintain high prediction accuracy over time.Limitations
Service Limits
LUIS has several limit areas, including model limits (e.g., number of intents, entities, and features), quota limits based on key type, and regional limitations. These limits can restrict the scope and complexity of the applications built using LUIS.Language and Feature Parity
Not all features in LUIS are at the same language parity. Performance can vary significantly across different features and languages, which may affect the overall usability and accuracy of the system.Quality of Input Text
The performance of LUIS is heavily dependent on the quality and formatting of the incoming text. Noisy or poorly formatted text can significantly degrade the system’s performance.Speech-to-Text Issues
LUIS may not perform well with speech-to-text inputs, especially in noisy environments or for users with speaking difficulties. This can lead to inaccuracies in intent and entity recognition.Areas for Improvement
Semantic Similarity
LUIS has limited knowledge of broader NLP aspects such as semantic similarity without explicit identification in examples. For better semantic understanding, users are recommended to migrate to Conversation Language Understanding.User Feedback Mechanism
Implementing a mechanism for end-users to report errors can help developers identify and correct inaccuracies, thereby improving the model over time. Given that LUIS is scheduled to be retired on October 1st, 2025, and new resource creation has been disabled since April 1st, 2023, it is recommended to migrate existing applications to conversational language understanding for continued support and improved capabilities. Microsoft Language Understanding (LUIS) - Pricing and Plans
The Pricing Structure for Microsoft’s Language Understanding (LUIS)
The pricing structure for Microsoft’s Language Understanding (LUIS) in Azure is structured into several tiers, each with its own set of features and pricing models.
Free Tier
The Free tier of LUIS offers limited but useful capabilities:
- Authoring: 1 million authoring transactions and 1,000 testing prediction transactions per month.
- Prediction: 10,000 prediction transactions per month. This tier is available for both web and container deployments.
Standard Tier
The Standard tier provides more extensive usage:
- Transactions: This tier is priced per 1,000 prediction transactions. The exact pricing varies based on the volume of transactions, with rates for 1 million, 5 million, and 25 million transactions per month.
- Throughput: The Standard tier supports up to 50 transactions per second (TPS).
Commitment Tiers
For larger usage, commitment tiers are available:
- These tiers offer pricing discounts for committing to higher volumes of transactions (e.g., 1 million, 5 million, and 25 million transactions per month). Overage rates apply for transactions beyond the committed volume.
Additional Features and Considerations
- Text and Speech Requests: Pricing applies to both text and speech requests, with separate rates for each type of request.
- Connected Containers: Similar pricing models are available for connected container deployments, with rates based on the number of transactions.
- Government and Special Agreements: Pricing can vary based on the type of agreement, such as those for US government entities or other special agreements. These entities may have different pricing structures and purchase options.
Pricing in Other Contexts
In some regions or through specific programs, the pricing may be presented differently. For example, through the UK’s G-Cloud framework, LUIS is priced at £1.52 per unit, though the definition of a ‘unit’ in this context would need clarification from the provider.
To get precise and up-to-date pricing, it is recommended to use the Azure pricing calculator or to contact an Azure sales specialist, as prices can vary based on the specific agreement and location.
Microsoft Language Understanding (LUIS) - Integration and Compatibility
Integration with Other Tools
LUIS has been widely used in conjunction with other Microsoft services and tools, particularly in the development of chatbots and other conversational AI applications.Azure Bot Service
LUIS can be integrated with Azure Bot Service to recognize intents and entities in user input. For example, in a flight booking application, LUIS helps the bot understand user requests and extract relevant information such as dates, destinations, and names.Azure Cognitive Services
LUIS is part of Azure Cognitive Services, which allows it to be used alongside other cognitive services like speech recognition, text analytics, and more.Custom Applications
LUIS can be integrated into custom applications using APIs and SDKs available in .NET, Python, Java, and Node.js. This allows developers to incorporate natural language understanding capabilities into their applications.Compatibility Across Platforms and Devices
Multi-Language Support
While LUIS itself does not support multilingual models natively, the new CLU offers multilingual support, allowing models to be trained in one language and predict intents and entities across 96 languages.Device Compatibility
LUIS can be integrated into various devices and platforms, including web applications, mobile apps, and IoT devices, as long as these devices can make API calls to the LUIS service.SDK Support
LUIS provides SDKs for several programming languages, ensuring compatibility with a wide range of development environments. However, with the migration to CLU, some refactoring will be necessary to use the CLU authoring and runtime APIs.Migration to Conversational Language Understanding (CLU)
Given the upcoming retirement of LUIS, it is recommended to migrate existing LUIS applications to CLU. CLU offers several advantages, including improved accuracy with state-of-the-art machine learning models, multilingual support, and easier integration with other projects using orchestration workflows.Backwards Compatibility
CLU provides backwards compatibility with previously created LUIS applications, making the migration process smoother. However, some features like `Pattern.Any` entities, entity roles, and certain settings will not be carried over during the migration.API and SDK Changes
The migration to CLU requires refactoring to use the new CLU authoring and runtime APIs, which are different from those used in LUIS.Role-Based Access Control (RBAC)
Both LUIS and CLU support Azure role-based access control (Azure RBAC) for managing access to resources. However, after migrating to CLU, the RBAC settings for Language resources must be manually added. In summary, while LUIS has been a powerful tool for integrating natural language understanding into various applications, its impending retirement necessitates a transition to CLU, which offers enhanced features and better compatibility across different platforms and languages. Microsoft Language Understanding (LUIS) - Customer Support and Resources
Support Options
Azure Portal Support
You can find solutions to common issues related to LUIS and other Azure AI services directly in the Azure portal. To do this, go to your Azure AI services resource, select Help in the left pane, and then choose Support Troubleshooting. Here, you can describe your issue and find relevant Learn articles and other resources to help resolve it.Create a Support Request
If you need more detailed support, you can create and manage support requests in the Azure portal. When submitting a support request, select Cognitive Services in the Service type dropdown field to ensure your issue is directed to the appropriate team.Microsoft Q&A and Stack Overflow
For quick and reliable answers to technical questions, you can engage with Microsoft Engineers, Azure Most Valuable Professionals (MVPs), or the expert community on Microsoft Q&A or Stack Overflow. Use relevant tags such as Language Understanding (LUIS), Azure AI Language, or Azure QnA Maker to categorize your questions.Additional Resources
Documentation and Guides
Microsoft provides extensive documentation for LUIS, including quickstarts, how-to guides, concepts, and tutorials. These resources help you get started with LUIS, manage your resources, and use the service in more specific or customized ways.LUIS Portal and Azure CLI
You can create and manage LUIS resources using the LUIS portal, Azure portal, or Azure CLI. The documentation guides you through the process of creating authoring and prediction resources, which are essential for training, testing, and deploying your LUIS applications.Migration to Conversational Language Understanding
Given that LUIS will be retired on October 1st, 2025, Microsoft recommends migrating your LUIS applications to conversational language understanding. This next-generation service offers state-of-the-art language models, multilingual capabilities, and better integration with other Azure services.Feedback and Feature Requests
If you have suggestions or need new features, you can post your feedback on the Azure feedback portal. This allows you to contribute to the improvement of Azure AI services, including LUIS and its successor, conversational language understanding. By leveraging these support options and resources, you can effectively manage and optimize your use of the LUIS service within the Azure AI ecosystem. Microsoft Language Understanding (LUIS) - Pros and Cons
Advantages of Microsoft Language Understanding (LUIS)
Simplicity and Ease of Use
LUIS is designed to be user-friendly, even for those without in-house AI expertise or prior machine learning knowledge. It allows users to build custom conversational AI applications with just a few clicks, using quickstarts or prebuilt domain apps.
Security, Privacy, and Compliance
LUIS is backed by Azure infrastructure, ensuring enterprise-grade security, privacy, and compliance. User data remains secure, encrypted in storage, and can be deleted at any time.
Integration Capabilities
LUIS integrates seamlessly with other Microsoft services such as the Microsoft Bot Framework, QnA Maker, and Speech service, making it easy to create sophisticated bots and other conversational AI solutions.
Custom Machine Learning
LUIS applies custom machine-learning intelligence to predict the overall meaning of conversational text and extract relevant, detailed information. It supports various entity types, including list, regex, and prebuilt entities.
Disadvantages of Microsoft Language Understanding (LUIS)
Upcoming Retirement
LUIS is scheduled to be retired on October 1, 2025. As of April 1, 2023, users cannot create new LUIS resources, making it essential to migrate existing applications to Conversational Language Understanding (CLU).
Limited Multilingual Support
Unlike CLU, LUIS does not support multilingual models. Each LUIS application is limited to a single culture, which can be a significant limitation for global applications.
Higher Data Requirements
LUIS requires more examples to generalize certain concepts in intents and entities compared to CLU, which uses state-of-the-art models that generalize better with less data.
Feature Limitations
Certain features like `Pattern.Any` entities, entity roles, and specific settings (e.g., normalize punctuation, normalize diacritics) are not supported or will be removed in the migration to CLU. Additionally, structured ML entities and phrase list features have limitations that are addressed differently in CLU.
Technical Adjustments Needed for Migration
Migrating from LUIS to CLU requires refactoring to use the new CLU authoring and runtime APIs, as well as adjustments to how entities and intents are handled. This can involve significant technical changes.
In summary, while LUIS offers simplicity, security, and integration capabilities, its upcoming retirement, limited multilingual support, and higher data requirements are significant drawbacks that users need to consider, especially when deciding whether to migrate to the more advanced Conversational Language Understanding (CLU) service.
Microsoft Language Understanding (LUIS) - Comparison with Competitors
When Comparing Microsoft’s Language Understanding (LUIS) with Other AI-Driven Analytics and NLP Tools
Unique Features of LUIS
- Simplicity and Accessibility: LUIS stands out for its ease of use, allowing developers to build custom language models without requiring extensive machine learning expertise. It offers prebuilt domain apps and quickstarts, making it accessible even for those new to NLP.
- Integration with Microsoft Ecosystem: LUIS seamlessly integrates with other Microsoft services such as the Microsoft Bot Framework, QnA Maker, and Speech service, which can be a significant advantage for businesses already using Microsoft tools.
- Security and Compliance: Backed by Azure infrastructure, LUIS provides enterprise-grade security, privacy, and compliance, ensuring data encryption and the ability to delete data at any time.
Potential Alternatives and Competitors
Google Cloud Natural Language API
- This API offers advanced NLP capabilities, including sentiment analysis, entity recognition, and text classification. It is highly scalable and can handle large volumes of text data. However, it may require more technical expertise compared to LUIS.
Salesforce Einstein Analytics
- Einstein Analytics is an AI-powered analytics platform that focuses on customer data analysis and predictive modeling. While it is strong in sales and customer relationship management (CRM) contexts, it may not offer the same level of custom NLP model building as LUIS. It is more specialized in analyzing customer behavior and preferences.
SAS Visual Analytics
- SAS Visual Analytics uses AI to automate data analysis and provide insights, including predictive models for customer behavior and sales trends. It is more focused on data visualization and exploration rather than pure NLP tasks, but it can handle natural language queries and provide AI-driven explanations of data patterns.
Tableau
- Tableau is a powerful data visualization and analytics platform that includes AI-powered features like natural language queries and predictive modeling. While it is excellent for data analysis and visualization, it does not specialize in NLP to the same extent as LUIS. However, its interactive dashboards and visualizations can be very useful for analyzing data derived from NLP tasks.
Future of LUIS
It’s important to note that LUIS is scheduled to be retired on October 1st, 2025, and new LUIS resources cannot be created after April 1st, 2023. Microsoft recommends migrating to the next generation of conversational language understanding, which offers state-of-the-art language models, multilingual support, and better integration with other Azure services.
In summary, while LUIS offers unique benefits in terms of simplicity, integration, and security, other tools like Google Cloud Natural Language API, Salesforce Einstein Analytics, SAS Visual Analytics, and Tableau provide different strengths and may be more suitable depending on the specific needs of the business. As LUIS is being phased out, considering the next generation of conversational language understanding or other alternatives is crucial for long-term support and functionality.
Microsoft Language Understanding (LUIS) - Frequently Asked Questions
Frequently Asked Questions about Microsoft’s Language Understanding (LUIS)
What is Language Understanding (LUIS)?
LUIS is a cloud-based conversational AI service that uses custom machine-learning intelligence to interpret user conversational, natural language text. It predicts the overall meaning and extracts relevant, detailed information from the text.How do I get started with LUIS?
To get started, you can sign in to the LUIS portal and create a new LUIS resource. You can choose to build your custom application from scratch or use one of the prebuilt domain apps. Follow the quickstart guides or how-to articles for more detailed instructions.What are intents and entities in LUIS?
Intents are names that represent the actions or goals a user wants to achieve, such as “BookFlight” or “OrderPizza.” Entities are the parameters needed to execute these actions, like the travel destination, date, or airline. For example, a “BookFlight” intent might require entities like the destination and travel date.How do I build and train a LUIS model?
You can build a LUIS model using the Authoring APIs or the LUIS.ai web app. Start by defining your intents and providing sample utterances for each intent. Label the relevant parts of the utterances as entities. Once the model is designed, train and publish it. The model can then receive and process user input via HTTP requests.Can I use prebuilt domain models in LUIS?
Yes, LUIS offers prebuilt domain models that include intents, utterances, and entities for common domains. These models are a great way to start using LUIS quickly, as they provide all the necessary components for your application.How does LUIS handle real user input and improve over time?
LUIS uses active learning to improve its performance over time. After your application is published, LUIS identifies real user utterances it is unsure about and provides them for you to review and label. This process helps LUIS learn and improve its accuracy with minimal effort from you.Can I integrate LUIS with other Microsoft services?
Yes, LUIS can be easily integrated with other Microsoft services such as the Microsoft Bot Framework, QnA Maker, and Speech service. This integration allows for more comprehensive and interactive conversational applications.What are the pricing tiers for LUIS resources?
LUIS offers different pricing tiers for authoring and prediction resources. The free (F0) tier provides 1 million authoring transactions and 1,000 prediction transactions per month for authoring, and 10,000 prediction endpoint requests per month for prediction. There is also a Standard (S0) tier for paid services.Is LUIS being retired, and what are the alternatives?
Yes, LUIS will be retired on October 1st, 2025, and new LUIS resources cannot be created after April 1st, 2023. Microsoft recommends migrating to conversational language understanding, which is the next generation of LUIS and offers improved features like multilingual support and state-of-the-art language models.How does conversational language understanding differ from LUIS?
Conversational language understanding is the next generation of LUIS, featuring transformer-based models, multilingual capabilities without retraining, and the ability to orchestrate between multiple language applications. It also integrates seamlessly with the Bot Framework SDK and offers comprehensive security and compliance.Can I use LUIS in multiple languages?
While LUIS itself does not support multilingual capabilities natively, the new conversational language understanding feature allows you to train models in one language and use them in multiple languages without retraining.How do I manage and monitor LUIS resources in Azure?
You can manage and monitor LUIS resources through the Azure portal. Here, you can view usage metrics, change pricing tiers, and customize resource usage charts to track your application’s performance. Microsoft Language Understanding (LUIS) - Conclusion and Recommendation
Microsoft’s Language Understanding (LUIS)
LUIS is a cloud-based conversational AI service that has been a significant tool for developers to build applications that can interpret and respond to natural language inputs. Here’s a final assessment of LUIS and its benefits, as well as guidance on who would benefit most from using it.
Key Features of LUIS
- Natural Language Processing (NLP): LUIS uses machine learning to analyze conversational text and predict the overall meaning, extracting relevant and detailed information.
- Ease of Use: LUIS is designed to be user-friendly, allowing developers to build custom language models without requiring in-house AI expertise. It offers prebuilt domain models and quickstarts to get started quickly.
- Integration: LUIS integrates seamlessly with other Microsoft services such as the Microsoft Bot Framework, QnA Maker, and Speech service, making it versatile for various applications.
- Security and Compliance: Backed by Azure infrastructure, LUIS ensures enterprise-grade security, privacy, and compliance, with data encryption and the ability to delete data at any time.
Who Would Benefit Most
- Developers: Developers looking to create conversational interfaces for applications such as chatbots, voice assistants, or any system that needs to interpret natural language inputs would greatly benefit from LUIS.
- Businesses: Enterprises in various sectors like banking, travel, entertainment, and more can use LUIS to build enterprise-grade conversational bots that enhance customer interaction and service.
- IoT Developers: Those working on IoT projects can leverage LUIS to create seamless conversational interfaces for controlling internet-accessible devices.
Migration and Future Support
- Retirement of LUIS: It is important to note that LUIS will be retired on October 1st, 2025, and new LUIS resources cannot be created after April 1st, 2023. Microsoft recommends migrating LUIS applications to the next generation of conversational language understanding to continue benefiting from product support and multilingual capabilities.
Recommendation
Given the impending retirement of LUIS, it is highly recommended to start planning the migration to the next generation of conversational language understanding. This new service offers state-of-the-art language models, multilingual support, and the ability to train in one language and use the model in multiple languages without retraining.
For those currently using LUIS, the transition to the new service will provide continued support and enhanced features, ensuring that your applications remain effective and up-to-date with the latest advancements in NLP and conversational AI. For new users, starting with the next generation of conversational language understanding will offer the best long-term benefits and support.