Welcome to our Customer Service Portal

Knowledge Base

How does Nanorep find an answer?

Nanorep maintains a knowledge base of questions and answers.  When a user asks a question, Nanorep looks for a similar question in the knowledge base. In case a similar question is found, the corresponding answer is then returned to the user.  If the answer is not in the knowledge base yet, the user can escalate it to an agent via ticket or chat.

Search Accuracy

Nanorep uses natural-language based search to analyze the meaning and the context of the question. During the search process, Nanorep looks for a question which has similar meaning and context.

Nanorep also uses the following tools for optimizing its search process:

  1. Nanorep utilizes a morphological engine which determines the source of the word (e.g., “selling” will be recognized as “sell”, “working as “work”). The engine has the capacity to handle complicated users expressions.
  2. Nanorep utilizes adaptive spelling correction capabilities in its search engine. This provides correct results even if they misspell their question. The spelling correction engine will adapt itself to your business language, based on the content of your knowledge base (i.e., product names and unique terms will be supported and understood by your Nanorep).

Every answer in the knowledge base can be related to multiple questions:

Users ask similar questions in various phrasings. Nanorep allows you to associate multiple questions (phrasings) to an answer.

For example, a knowledge base which has an answer to the question "Do you sell fruit?" might have the following questions associated with this question:

  1.  "Do you sell Apples?"
  2.  "Can I buy Bananas?"
  3.  "Can I purchase an Orange?"

Accordingly, when the question “do you sell bananas?" is asked, the system will perform the following actions:

  1. Searching for a similar question in the knowledge base.
  2. Recognizing an existing similar question (phrasing) in the knowledge base, such as "Can I buy Bananas?"
  3. Sending the associated answer back to the user.