Translated from 日本語 • View original In previous articles, we covered Azure Cognitive Services in general. In this article, I would like to introduce the language field service, the content of the service, and the billing system that you are interested in.
What are Azure Congnitive Services language disciplines?
As previously introduced, in the language field, you can mainly incorporate the function of understanding (analyzing) the content of a sentence into the service. In the language field, you can understand the meaning of the transcribed sentence and connect it to the next action.
You can:
- Understand the message entered by the user in the chatbot and continue the conversation
- Enter a sentence and pick up important words in it
- Translate large volumes of documents
Click here for the official website of the language field
Services in the Field of Languages
There are five services in the field of speech.
Entity recognition
Entity recognition can extract keywords from text, such as dates or specific information about products. For example, you can find dates in sentences that are expressed in various ways, such as "tomorrow", "the day after tomorrow", "August 9th", "8/9", etc. You can use this feature to automatically create a meeting call in a chat exchange, for example, or to extract the appointment date in a chat-based booking system. Entity recognition can also be used in environments without an Internet connection (separate application required).
Billing System
The billing system for Entity recognition is pay-as-you-go and pre-booking for what you use. You define one text record for one sentence (up to 1000 characters) and are charged per text record. With pay-as-you-go billing, you are charged for every 1000 text records. For pre-bookings, a flat rate is charged up to a minimum of 1 million text records, followed by an additional fee for each 100 text records. You can increase the number of quotas available at a fixed rate according to the scale of use.
Sentiment analysis
Sentiment analysis is a service that allows you to evaluate and analyze opinions and emotions from writing. You can analyze the reputation of your products and brands posted on social media (SNS) and customer reviews posted on your e-commerce site to automatically determine whether they are positive or negative about the product or your company. Positive or negative indicators are expressed in numbers, so you can efficiently collect especially positive and negative opinions and use them for the next time. Combined with the Entity recognition mentioned earlier, you can list the positive and reputable features. Sentiment analysis can also be used in environments without an internet connection (separate application required).
Billing System
The billing system of Sentiment analysis is pay-as-you-go billing and pre-booking for what you use. You define one text record for one sentence (up to 1000 characters) and are charged per text record. With pay-as-you-go billing, you are charged for every 1000 text records. For pre-bookings, a flat rate is charged up to a minimum of 1 million text records, followed by an additional fee for each 100 text records. You can increase the number of quotas available at a fixed rate according to the scale of use.
Question Answering (formerly QnA Maker)
Question Answering is a question-and-answer service. Find and output the pre-set answer for the sentence in the input. It is an evolution of a service that was previously called QnA Maker. For example, you can connect MS Teams chatbots to create FAQ bots. Click here for the official website of Question Answering
Billing System
The billing system of Question Answering is pay-as-you-go for what you use. You define one text record for one sentence (up to 1000 characters) and are charged per text record. With pay-as-you-go billing, you are charged for every 1000 text records.
However, Question Answering incurs a separate server fee to use this feature.
Conversational language understanding (formerly LUIS)
Conversational language understanding can find the "intent" of the entire sentence and the important word "entity" in the sentence. It is an evolution of a service that was previously called LUIS.
For example, in a conversation that determines a meeting, identify the following intents:
- Remarks about "meeting room location"
- Remarks on "meeting time"
- Remarks about "meeting participants"
Then, within that statement, you can extract key entities related to the intent, such as:
- Is the location "Conference Room A"?
- Is the time "from 10:00"?
- Do you call "Hotta" to the meeting?
By combining this service with Question Answering, you can create advanced chatbots. Like the previous example, a meeting coordination bot and a redelivery reception bot. In addition, you can create home appliance operation services by linking with voice assistants such as Google Home and Alexa.
Click here for the official website of Conversational language understanding
Billing System
The billing system of Conversational language understanding is pay-as-you-go for what you use. You define one text record for one sentence (up to 1000 characters) and are charged per text record. With pay-as-you-go billing, you are charged for every 1000 text records.
However, Conversational language understanding incurs a separate server fee to use this feature.
Translator
Translator is a text translation service. When you enter text data, you can identify which verbalization and translate it into any language. It also allows for real-time translation, so you can translate conversations on the fly by combining it with Speech To Text, which I introduced last time. For example, you can use it when you want to translate a large amount of text at once.
Translator can also be used in environments without an internet connection (separate application required).
Billing System
Translator's billing system is pay-as-you-go and pre-booking for what you use. With pay-as-you-go billing, you are charged for every million characters translated. If you pre-order, you will pay a flat rate of up to 250 million characters, and then you will be charged an additional fee for every 1 million characters thereafter. Depending on the expected scale of use, you can use it differently by pay-as-you-go or advance reservation.
Conclusion
In this article, we introduced the language field of Azure Cognitive Services. The billing system for each service is pay-as-you-go, but if you store files for use in the cloud, you will be charged separately for storage. Colorkrew will provide support in various ways, from PoC to main development using Azure Cognitive Services. In the next article, we'll introduce you to the visual field of Azure Cognitive Services. If you are interested in developing assistance with Azure Cognitive Services, please contact us at:
Cloud Utilization Support Service | Kuramane
