Search

The search endpoint lets you perform search on transcripts of Conversation objects associated with your account. There are three types of searches you can perform in Sift.

Standard search is the default search mode for Sift. This mode computes search rankings based on word matchings between the search query and the transcript.

Phonetic search uses the double metaphone phonetic encoding algorithm to perform searches based on the pronounciation of words. For example, the search query “fair” would match with any appearances of “fare” in transcripts.

Semantic search is fuzzy search mode that matches excerpts in transcripts with similar semantic contexts. For example, the search queries “canine” and “cat” would match with the word “dog” in a transcript.

Search Conversations

Request

url:https://api.gridspace.com/v0/search method:POST

Property	Type	Description
At Least One Required Properties
query	string	The query you want to search for in the audio and conversation name.
missing_query	string	The words you want to be missing from the audio and conversation name.
conversation_ids	list	A list of ids of Conversation objects to search in. If this parameter is not supplied, then we search through all conversations.
tags	list	A list of conversation tags.
before	string	Find Conversation that have a start time before the provided value. ISO 8601 formatted date and time.
after	string	Find Conversation that have a start time after the provided value. ISO 8601 formatted date and time.
Optional Properties
is_phonetic	boolean	Whether or not to enable phonetic search for searching the audio only. Defaults to False. (Note: turning on phonetic search could drastically increase the number of false matches)
is_semantic	boolean	Whether or not to enable semantic search for searching the audio only. Defaults to False. (Note: turning on semantic search could drastically increase the number of false matches)
offset	int	The start position of the search results. Defaults to 0.
size	int	The number of search results to return. Defaults to 10.

Response

content-type:application/json

The response body contains a JSON object representing a paginated list of search results. See Paginated Results for more information. They are sorted by their relavance score if a query is provided. Otherwise they are sorted on the start time of the conversation with the most recent one first.

Property	Type	Description
Required Properties
next	string	URL of the next page of results, or null if this is the last page.
previous	string	URL of the previous page of results, or null if this is the first page.
total	int	The total number of search hits.
results	list	A list of search results.
results[]. conversation_id	string	The id of a conversation object where at least one search result is found.
results[].name	string	Name of Conversation.
results[].start_time	string	ISO 8601 formatted date and time.
results[].transcript	string
results[].tags	list	List of tags.
results[].tags.tag	string	A tag that was applied to the Conversation.
results[]. matched_transcript	list	List of dictionaries.
results[]. matched_transcript. index	int	The index of the word in the transcript.
results[]. matched_transcript. word	string	The word that was matched in the transcript.
results[]. matched_transcript. start_ms	int	The start time of the word in the audio file. Units are milliseconds.

Example

{
  "next" : null,
  "previous" : null,
  "total" : 2,
  "hits" : [
    {
      "conversation_id" : "2af22188c5c97256",
      "total" : 1,
      "hits" : [
        {
          "excerpt" : "this was searched",
          "start_ms" : 230,
          "duration_ms" : 2420
        }
      ]
    },
    {
      "conversation_id" : "5d3dc73fc8cf34be",
      "total" : 1,
      "hits" : [
        {
          "excerpt" : "another search query",
          "start_ms" : 28220,
          "duration_ms" : 30020,
        }
      ]
    }
  ]
}