Builder Guidelines
How to craft utterances that will pass the validation process
Last updated
How to craft utterances that will pass the validation process
Last updated
Connect to the Platform
In the 'getting started' module, we explained how to connect to the Synesis One AI training platform. Let's do that now. Go to Synesis.one, connect your wallet, and select the builder role. For illustration purposes, we'll show you how to do train2earn on your desktop, but its basically the same for smartphone.
Select a Campaign
Next, select a campaign (retail, telecommunications, etc.) from the list of available projects. You stake 100 SNS for access. These funds will be returned to you when the campaign ends (typically one week). Select 'approve' when the popup screen asks you to confirm the transaction.
Submit Utterances
Let's say you select a real estate campaign. The architect who created the campaign wants to train a chatbot to understand customer queries about how to take possession and move into a new apartment. Your task is to think of all the ways this might be expressed and to submit them as utterances. Before you begin, be sure to read the campaign description and study the example sentences, as well as the utterances that other builders have already submitted.
You can submit up to 50 utterances at a time. Note the system will reject duplicate utterances and prompt you to try a different sentence. When you've reached the 50 utterance limit, click the 'submit utterances' button and your submission will be recorded on the blockchain. This can take a moment.
Utterance Types
Each campaign features a topic sentence. In this case 'I want to know about the turnover of my unit.' Beneath (left side) you'll see three selectable categories:
Find a more specific sentence
Find a more general sentence
Find an entailment for the sentence
As explained above, these three categories are rooted in linguistic theory but also map on to Mind AI’s unique three-node data structure (called canonicals) that the engine uses to establish logical relations between ontologies). Let's go over the three kind of utterances.
Specific Utterances
A specific utterance, as the name suggests, is one that is related to the topic of the campaign, but with more detail — such as details of an action, a situation, or a specific intent. Typically, this means using more words, which makes the utterance longer. You can do this by using more detailed words than the original topic or changing the sentence structure (syntax) to express the same idea. Consider the following example in the domain of electronics:
In the above utterances, the underlined word(s) provide greater specificity to the topic installation services availability. These utterances teach the AI different ways that someone might inquire about installation services, which will enable the AI to understand customer inquiries on this topic in the future. Note that the validators will reject utterances with the same sentence structure but different words for the same thing. So, for example, ‘Do you offer installation services for television / stereos / satellite dishes / home theater / etc. will be rejected, as they don’t teach the AI anything new.
The example of topic subject 2 that will be rejected by validators:
✘ Is there a lady parking lot?
(Note: in many Asian countries, there are parking spaces or lots reserved for women drivers). This sentence may look like a specific sentence at first glance, but the content does not cover the meaning of the topic subject (“at the branch” is missing). This sentence becomes more general to be categorized as specific.
General Utterances
A general utterance, as you might guess, refers to the theme described in the topic subject example, but with fewer details. General utterances will include less details of action, situation, and specific intent, thus making them shorter.
If the utterances are too broad and fall outside the scope of a given domain (in this case, telecommunications), then they will be rejected by the validators. Consider the following:
✘ Can I buy a package plan?
✘ I am interested in buying a package plan
These utterances do not describe “mobile package plan” in a more general sense, but rather invite different interpretations in different domains (such as travel package plan).
The examples of topic subject 2 that will be rejected by validators:
✘ Can I drive to the store?
✘ I will go to the branch by car
These sentences do not clearly describe the speaker’s intent to know whether a parking lot is available at the branch. They can be validated as entailments, but not general sentences of topic subject 2.
The example of topic subject 3 that will be rejected by validators:
✘ I want to make a payment
✘ Cash on delivery
The first sentence is too broad. It does not describe “cash” or “delivery” therefore changes the speaker’s intent. The second example is too broad and ambiguous. It can refer to “I want to know about cash payment on delivery”, “I have a problem with cash payment on delivery” or anything about cash on delivery.
Entailment
Entailment describes a relationship between two sentences such that if the first one is true, the second must also be true. Entailment allows the AI to understand and reason logically based on natural language inputs, thereby improving the AI’s ability to make its own hypotheses about a given topic. Consider the following example from the Food & Beverage domain:
(1) Topic Subject: May I have some water?
(2) Entailment Utterance 1: I am thirsty.
(3) Entailment Utterance 2: I need water.
In this case, (1) entails the meaning of both (2) and (3). We can say that the Topic Subject entails the meaning of Entailment Utterances. Builders should create utterances that can replace the Topic Subject and keep the original intent intact.
The following utterance does not fulfill the topic subject’s intent though it is entailed by the topic subject.
✘ People can misunderstand how to use a laptop
To be validated, utterances should fulfill the intent of the Topic Subject, meaning they can replace the Topic Subject and still convey the same meaning. Given that the Topic Subject is true, utterances must also be true. Finally, entailment utterances should capture a part of information from the Topic Subject.
The example of topic subject 2 that will be rejected by validators:
✘ Your company has put forward a technical white paper in the past
✘ I’m interested in technology
These sentences do not properly fulfill the topic subject’s intent though it is entailed by the topic subject.
The example of topic subject 3 that will be rejected by validators:
✘ People can apply for a housing loan
✘ I don’t have enough money to buy a house
These sentences do not properly fulfill the topic subject’s intent though it is entailed by the topic subject.
If you become a skilled builder, you may be invited to try out for a Validator Role.
