Create Artifacts using Templates

Luma Knowledge provides predefined Templates that can be used by a Curator to generate Artifacts. These define the artifact type and required fields. Using these Templates, the Curator can create Knowledge Artifacts manually by adding information content.

There are two types of templates available in Luma Knowledge:

1. Regular Template

2. FAQ Template

Regular Template

Regular Template allows a Curator to use the organization-specific templates to create Artifact. Organization-specific custom templates enable the curator to create artifacts using documents that do not follow a specific format. These documents can be used to create the artifacts manually or import documents using these templates.

For Tenants with Metadata Search:

For tenants with Metadata Search, Artifact creation using Regular Templates includes the following steps:

Open

1. Adding Artifact details- The curator selects an organization-specific Knowledge Template and adds the content. The selected template determines the mandatory and optional artifact fields and the information to be used to create an ontology.

2. Create Artifact- Once the data is filled, click on Create Artifact. This initiates Artifact creation. For Artifact using Regular templates, the Semantic Analysis process is not initiated; however, the system does generate Ontology and other mandatory details. These are created based on your tenant-level configurations.

3. Ontology/ Keyphrase generation- Based on the Artifact content, the system generates Key phrases and other keywords. These are used to identify matching artifacts during the search.
   Luma Knowledge uses Generative AI to generate Ontology. If Large Language Models such as Open AI are enabled for your tenant, the system will automatically extract relevant keywords from documents and data, making information retrieval faster and more effective. The system can generate valid keywords and Key phrases reducing the time and effort curators spend in adding and maintaining Metadata for Knowledge Artifacts.

4. Summarization: It is important that the Summary is well-formed and articulates the information in the document well. Artifact’s Summary is used to generate Metadata. An end-user could manually add the Summary for the artifact. Alternatively, Luma Knowledge can use Generative AI to Summarize Knowledge Articles. If Large Language Models such as Open AI is enabled for the tenant, the curator does not need to add a summary for each artifact. Luma Knowledge automatically generates the summary based on the artifact content.

5. Question and Answer Generation: The feature is available when Large Language Models such as Open AI are enabled for your tenant. Using Open AI, Question and answer pairs (QnApairs) is generated and recorded as part of the artifact. These QnA pairs enable Luma Knowledge to present precise information to the end-users quickly, reducing the time spent on answering customer inquiries.

For Tenants with Semantic Search:

If Semantic Search is enabled for the tenant, Ontology generation, Summarization, and FAQ generation is NOT required. Artifact creation includes the following steps:

Open

1. Add Artifact Details: The curator selects an organization-specific Knowledge Template and adds the content. The selected template determines the mandatory and optional artifact fields and the information to be used to create an ontology.

2. Create Artifact: Once the data is filled, click on Create Artifact. This initiates Artifact creation. For Artifact using Regular templates, the Semantic Analysis process is not initiated; however, the system does generate Ontology and other mandatory details. These are created based on your tenant-level configurations.

3. Data Extraction and Chunking: Once the artifact is created, text is extracted from the source. The extracted text is then divided into manageable chunks, typically at the semantic level (e.g., paragraphs). This chunking facilitates efficient processing and retrieval.

4. Vectorization: The text chunks undergo embedding, where they are converted into vectors or numerical forms. This conversion is crucial for enabling efficient and accurate search capabilities. The resulting embeddings are stored in a vector database, ready for search and retrieval operations.

FAQ Template

FAQ Template allows a Curator to create an Artifact and add Question and Answer (Q&A) pairs to hold knowledge. For each Q&A pair, Ontology analysis is performed, generating metadata for the content as well as Q&A pairs. These Q&A pairs linked to the Artifact as well as the topic identified during Ontology generation.

Using the FAQ template, you can manually create an Artifact and add Questions and Answers pairs. Alternatively, you can also upload an excel file with Questions and Answers.

Artifact creation using FAQ Templates includes the following steps:

1. Add Artifact Details: The curator selects the FAQ Template and adds the content. Details just as Artifact Name and Domain are selected.

2. Add summary: The system uses the artifact Summary to generate Metadata, which is in turn used to identify relevant artifacts for a user’s search. It is essential that the Summary is well-formed and articulates the information in the document well. When creating an artifact, the curator adds the summary manually. The system then uses the summary to create the Metadata/Ontology.

3. Add QnA Pairs: In these Artifacts, the Knowledge is added in the form of QnA pairs. These pairs are not automatically generated but are added manually by the user.