Knowledge Center

Parquet data dictionary for customer hosting

A deep-dive into the parquet file and its data dictionary

Overview

In this section we dive into the actual setup for data delivery into your cloud storage system. We use Parquet as our intermediate data format to pass from Analytics studio to the BI platform of your choice. We choose Parquet as it is a columnar data format that is efficient, scalable and supports analytics. It stores data organized by column which makes it suitable for analytics systems. It supports compression and encoding techniques to reduce the file size. It supports schema evolution, making it suitable as an intermediate data format that can be passed into different analytics platforms.

Table & Schema Design

It is often the case that customers want to view certain metrics or a subset of fields and not the list of observations. Our parquet table schema allows for this 'cherry pick' selection of fields based on the unique use-case of our customers analytics needs.

In the sections that follow, we will go over the purpose of each table and how their fields link with each other as well as their real world meaning.

Tables

Content

Call level summary
Content Metric Transposed

KPIs and other 40+ paralinguistic metrics about the conversation itself
Custom Content

Custom content and metrics derived from media metadata non-paralingusitic and factual such as agent name, phone number, etc
Ontology

Flattens the multi-level ONTOLOGY - this is used in tableau joins.
Segment

Turn-level information about what was spoken and how it was spoken
Category Observation

Information of when a category mapping was detected within a call/message interaction
Verb Noun Observation

Verb-Noun pairs extracted from an interaction. Great for quick summary analysis.
Metric Observation

Information about measured conversation metrics.
Sentiment Observation

Information about calculated sentiment scores on turn-level data (stereo). If no sentiment was calculated there should be no entry. If mono calls, this table will not be as useful.
Entity Observation

Information about extracted entities
Keyword Observation

Information about extracted keywords discovered with semantic indexing. This data is not as useful for contact center calls.
Topic Observation

Information about extracted topics discovered with semantic indexing. This data is not as useful for contact center calls.
Conversation Survey

Information about all the surveys like questions, answers that have taken place.
Deleted Content

Content Marked for deletion.
Engagement Agent Participants

Information about all the agent related to engagements
Engagement Campaign Info

Information about all the campaign of the engagements.
Engagement Customer Info

Information about all the customers that have engaged.
Engagement Info

Information about all the Conversations that have happened
Intent Observation

Information about something that get the system observed happened during a conversation
Skill Transfers

Information about skill based transfers of conversations
Classifier Observation

Information about classifier results. If no classifier was used there should be no entry.
Engagement Attachments

Information about files and other attachments uploaded or linked in the interaction.
Engagement Purchase

Information about purchases that took place as part of the interaction.
Engagement Purchase Item

Information about the purchase items themselves in conjunction to the purchase data.
Detector Observation

Information about detectors that fired and their scores. If no detector fired there should be no entry and hence no parquet file either.
Agent Surveys

Agent Survey as filled in by the agent either during or after the conversation has ended, and lets the agent record things like the nature or outcome of the conversation.
Agent Survey Questions

Information about the questions marked against a survey.
Agent Survey Answers

Information about the answer provided within a survey.
Agent Status Event

Information about change in agent's status ie. Online, offline, away and so on.

Content

This is the most foundational table based on which almost all other custom metrics and fields are generated. This table contains the call level summary about the media file itself. The media file being the voice or text conversation that took place between the agent/bot and user.

Table Name	Primary Key	Join Keys
Content	Content ID	Content ID

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
media_id	VoiceBase mediaId - Unique id per call for VB API	CHAR(36)
account_id	VoiceBase (platform) account ID	CHAR(36)
account_name	Account name, whatever we use now, probably platform account name	VARCHAR(128)
title	media title	VARCHAR(255)
external_id	external media ID	VARCHAR(255)
language	Overall language of the interaction as pre-selected by client/customer.	VARCHAR(5)
start_datetime	Metadata from Customer - Call start date/time	DECIMAL(19,0)
end_datetime	end time we have now but in UNIX time (seconds since 1970/01/01)	DECIMAL(19,0)
duration_milli_sec	duration of audio file in milliseconds	DECIMAL(19,0)
redacted	A boolean field, it signals if any information at all was redacted from the interaction. EG: False if no informatio was deleted from the voice/message interaction. This comes in handy when you would simply and quickly want to know in summary how many calls had any senstive data mentioned. Instead of having to look through all instances within the audio/message you can simply use this flag.	BOOLEAN
interaction_type	Speech if voice, text if messaging.	VARCHAR(32)
num_audio_channels	Count the distinct number of non-blank/empty w values in the words array where m equals TURN.	DECIMAL(2,0)
upload_version	version from call details	VARCHAR(64)
media_content_type	content type of media file	VARCHAR(48)
transcript_status	Final status of VoiceBase transcription processing; finished, failed.	VARCHAR(32)
num_words	new field, number of words in the transcript (count all entries in the word array of type “w”)	INTEGER
metadata_json	Full JSON results	VARCHAR(8192)
created_datetime	UNIX timestamp when audio was received by VoiceBase	DECIMAL(19,0)
finished_datetime	UNIX timestamp when the VoiceBase system finished processing the audio	DECIMAL(19,0)
processed_datetime	UNIX timestamp when export files were created	DECIMAL(19,0)
publisher_vendor	Source of data (MIA / EH API) which API was used to retrive this data. For voice calls it is always null which is the default. In the near future it will include VoiceCore which is an additional engine.	VARCHAR(96)
source_system	Overall source of the data. I.E. either VoiceBase or LivePerson. Not particularly useful to customer analytics.	VARCHAR(96)
site_id	Unique numeric ID associated with LivePerson/Analytics Studio account.	VARCHAR(32)

Content Metric Transposed

Conversation metrics, KPIs and other 40+ bits of insights related to the interaction from the content observation table. You may think of this table as holding insights and key parameters about the call/messaging interaction itself. This includes (but not limited to) overall sentiment of the call, duration and instances of silence, amplitude and volume of sound within the call and so on.

Table Name	Primary Key	Join Keys
Content Metric Transposed	Content ID	Content ID

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
media_id	VoiceBase mediaId - Unique id per call for VB API	CHAR(36)
account_id	VoiceBase (platform) account ID	CHAR(36)
account_name	Account name, whatever we use now, probably platform account name	VARCHAR(128)
title	media title	VARCHAR(255)
external_id	external media ID	VARCHAR(255)
language	Overall language of the interaction as pre-selected by client/customer.	VARCHAR(5)
start_datetime	Metadata from Customer - Call start date/time	DECIMAL(19,0)
end_datetime	end time we have now but in UNIX time (seconds since 1970/01/01)	DECIMAL(19,0)
duration_milli_sec	duration of audio file in milliseconds	DECIMAL(19,0)
redacted	A boolean field, it signals if any information at all was redacted from the interaction. EG: False if no informatio was deleted from the voice/message interaction. This comes in handy when you would simply and quickly want to know in summary how many calls had any senstive data mentioned. Instead of having to look through all instances within the audio/message you can simply use this flag.	BOOLEAN
interaction_type	Speech if voice, text if messaging.	VARCHAR(32)
num_audio_channels	Count the distinct number of non-blank/empty w values in the words array where m equals TURN.	DECIMAL(2,0)
upload_version	version from call details	VARCHAR(64)
media_content_type	content type of media file	VARCHAR(48)
transcript_status	Final status of VoiceBase transcription processing; finished, failed.	VARCHAR(32)
num_words	new field, number of words in the transcript (count all entries in the word array of type “w”)	INTEGER
metadata_json	Full JSON results	VARCHAR(8192)
created_datetime	UNIX timestamp when audio was received by VoiceBase	DECIMAL(19,0)
finished_datetime	UNIX timestamp when the VoiceBase system finished processing the audio	DECIMAL(19,0)
processed_datetime	UNIX timestamp when export files were created	DECIMAL(19,0)
publisher_vendor	Source of data (MIA / EH API) which API was used to retrive this data. For voice calls it is always null which is the default. In the near future it will include VoiceCore which is an additional engine.	VARCHAR(96)
source_system	Overall source of the data. I.E. either VoiceBase or LivePerson. Not particularly useful to customer analytics.	VARCHAR(96)
site_id	Unique numeric ID associated with LivePerson/Analytics Studio account.	VARCHAR(32)

Custom Content

This table contains custom data derived from media metadata that is non-paralinguistic (unlike Content Metric Transposed) in nature. It contains factual meta-data about the call such as the name of the agent on call, their phone number, timezone, email, the name and phone number of their superiors and so on.

Table Name	Primary Key	Join Keys
Custom Content	Content ID

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for Analytics Studio	DECIMAL(19,0)
time_zone	This is a custom field open to any data input from the customer side.	VARCHAR(48)
source_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_ext_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_full_name	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_phone_number	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_email	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_level	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_location	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_score1	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_score2	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_team_ext_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_team_label	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_tenure	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_tz	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_manager_ext_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_manager_full_name	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_manager_phone_number	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_manager_email	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_supervisor_ext_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_supervisor_full_name	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_supervisor_phone_number	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_supervisor_email	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_coach_ext_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_coach_name	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_org_ext_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
agent_org_name	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_ext_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_full_name	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_phone_number	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_email	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_affiliation_dt	This is a custom field open to any data input from the customer side.	BIGINT
person_area_code	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_city	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_country	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_postal_code	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_state	This is a custom field open to any data input from the customer side.	VARCHAR(64)
person_tz	This is a custom field open to any data input from the customer side.	VARCHAR(64)
affiliated_person_name1	This is a custom field open to any data input from the customer side.	VARCHAR(64)
affiliated_person_name2	This is a custom field open to any data input from the customer side.	VARCHAR(64)
affiliated_person_role1	This is a custom field open to any data input from the customer side.	VARCHAR(64)
affiliated_person_role2	This is a custom field open to any data input from the customer side.	VARCHAR(64)
after_contact_work_duration	This is a custom field open to any data input from the customer side.	INTEGER
after_contact_work_end_dt	This is a custom field open to any data input from the customer side.	BIGINT
after_contact_work_start_dt	This is a custom field open to any data input from the customer side.	BIGINT
call_direction	This is a custom field open to any data input from the customer side particular to their unique use case. In this case: inbound or outbound (agnet called or customer call) sales call / query call	VARCHAR(32)
call_type	This is a custom field open to any data input from the customer side particular to their unique use case. In this case: support call / sales call / etc	VARCHAR(48)
fcr	This is a custom field open to any data input from the customer side particular to their unique use case. In this case: whether or not the customer's issue was resovled in the first call itself.	BOOLEAN
caller_location	This is a custom field open to any data input from the customer side.	VARCHAR(64)
campaign_name	This is a custom field open to any data input from the customer side.	VARCHAR(64)
campaign_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
codec_change	This is a custom field open to any data input from the customer side.	VARCHAR(64)
dialed_extension	This is a custom field open to any data input from the customer side.	VARCHAR(64)
dtmf_events	This is a custom field open to any data input from the customer side.	VARCHAR(64)
hold_events	This is a custom field open to any data input from the customer side.	VARCHAR(64)
hold_time	This is a custom field open to any data input from the customer side.	INTEGER
outcome_code1	This is a custom field open to any data input from the customer side.	VARCHAR(64)
outcome_code2	This is a custom field open to any data input from the customer side.	VARCHAR(64)
outcome_code3	This is a custom field open to any data input from the customer side.	VARCHAR(64)
incident_id	This is a custom field open to any data input from the customer side.	VARCHAR(64)
line_type	This is a custom field open to any data input from the customer side.	VARCHAR(64)
mac_address	This is a custom field open to any data input from the customer side. In this case it is an instance of a mac address.	VARCHAR(64)
num_evals	This is a custom field open to any data input from the customer side.	INTEGER
num_holds	This is a custom field open to any data input from the customer side. In this case it is an instance of number of holds while on the call	DECIMAL(3,0)
queue_duration	This is a custom field open to any data input from the customer side. In this case it is an instance of total wait time / queue duration	INTEGER
rx_codec	This is a custom field open to any data input from the customer side. In this case it is an instance of a codec they would like to have mentioned.	VARCHAR(64)
reason_code1	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
reason_code2	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
reason_code3	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code1	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code2	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code3	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code4	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code5	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code6	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code7	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
routing_code8	This is a custom field open to any data input from the customer side. In this case it is an instance of a routing code (as seen in dtmf systems) they would like to have mentioned.	VARCHAR(64)
survey_code1	This is a custom field open to any data input from the customer side. In this case it is an instance of a survey or campaign code they would like to mention	VARCHAR(64)
survey_code2	This is a custom field open to any data input from the customer side. In this case it is an instance of a survey or campaign code they would like to mention	VARCHAR(64)
survey_code3	This is a custom field open to any data input from the customer side. In this case it is an instance of a survey or campaign code they would like to mention	VARCHAR(64)
date_time1	This is a custom field open to any data input from the customer side. In this case it is one instance of a datetime figure a customer would like to mention.	BIGINT
date_time2	This is a custom field open to any data input from the customer side. In this case it is one instance of a datetime figure a customer would like to mention.	BIGINT
date_time3	This is a custom field open to any data input from the customer. In this case it is one instance of a numeric figure a customer would like to mention.	BIGINT
numeric_4_0	This is a custom field open to any data input from the customer. In this case it is an instance of any numeric figure the customer would like to mention.	NUMERIC(4,0)
numeric_10_0	This is a custom field open to any data input from the customer. In this case it is an instance of any numeric figure the customer would like to mention.	NUMERIC(10,0)
numeric_18_0	This is a custom field open to any data input from the customer. In this case it is an instance of any numeric figure the customer would like to mention.	NUMERIC(18,0)
numeric_4_3	This is a custom field open to any data input from the customer. In this case it is an instance of any numeric figure the customer would like to mention.	NUMERIC(4,3)
numeric_10_2	This is a custom field open to any data input from the customer. In this case it is an instance of any numeric figure the customer would like to mention.	NUMERIC(10,2)
numeric_18_8	This is a custom field open to any data input from the customer. In this case it is an instance of any numeric figure the customer would like to mention.	NUMERIC(18,8)
numeric_double	This is a custom field open to any data input from the customer. In this case it is an instance of any numeric figure the customer would like to mention.	DOUBLE

Segment

Detailed summary insights of various sections within a messaging or voice interaction. For example, if certain words were mentioned within a call, this table mentions the particular section (start time, end time / message count) where said word was mentioned.

TABLE_NAME	PRIMARY KEY	JOIN KEYS
Segment	segment_id	segment_id, content_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
segment_id	Should use a hash (MurrmurHash), possibly of mediaId+seq_num	DECIMAL(19,0)
content_id	Unique id of media file	DECIMAL(19,0)
segment_type	Turn level. Here a turn can be defined as all text sent by an agent until one is sent by the customer or vice versa.	VARCHAR(24)
seq_num	Sequencing number of segment order through call	SMALLINT
start_ms	Start milliseconds of segment	DECIMAL(19,0)
end_ms	End milliseconds of segment	DECIMAL(19,0)
duration_ms	Duration in milliseconds of segment	DECIMAL(19,0)
start_ratio	START_MS/Total length of the call	DECIMAL(5,3)
avg_word_confidence	Transcription confidence score within segment	DECIMAL(4, 3)
volume_avg	Average volume of segment	DECIMAL(10, 3)
volume_max	max volume within a segment	DECIMAL(10, 3)
volume_stdev	standard deviation of volume within a segment	DECIMAL(10, 3)
num_words	number of words within a segment	SMALLINT
num_chars	number of characters within a segment	INTEGER
conversation_sentiment	Sentiment value for this segment	DECIMAL(4, 3)
speaker_name	Name of speaker if provided	VARCHAR(48)
conversation_text	verbatim text of segment	VARCHAR(2000000)

Ontology

This table flattens the multi-level hierarchical grouping of categories into a single string for easier manipulation and plotting in BI platforms or general analytics. Think of it as a category of categories to help group data in a more summarised manner.

TABLE_NAME	PRIMARY KEY	JOIN KEYS
Ontology	ontology_id	ontology_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
ontology_id	Unique idenntifier for all ontologies	BIGINT
ontology_label	Name of ontology (e.g. category name)	VARCHAR(128)
observation_type	Type of observation (e.g. 'CATEGORY', 'DETECTOR',...)	VARCHAR(48)
collection_name	Highest level of categories. Collection_Group_Label ex: Call Driver, Agent Quality, Alerts	VARCHAR(128)
group_name	Sub groups of category organization ex: CallDriver_AccessIssue_CantLogin, Group=Access Issue	VARCHAR(64)
deleted	indicates whether an ontology was deleted from the VoiceBase Platform system. Boolean.	BOOLEAN

VerbNoun Observation

This table contains Verb-Noun pairs extracted from a conversation. These are great for quick, intuitive summary analysis of the purpose or outcome of a message or voice interaction.

Eg: Verb:Pay, Noun:Bill -- This intuitively tells us that the interaction has something to do with the user wanting to money and or paying bills.

Table Name	Primary Key	Join Keys
VerbNoun Observation	Obvesrvation ID	Content_Id, Segment_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Generated reproducible hash code, MurmurHash(mediaId+verb+”/” noun)	DECIMAL(19,0)
observation_type	“VERBNOUN”	VARCHAR(48)
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
segment_id	Reference to segment if there is one	DECIMAL(19,0)
verb	This is the verb extracted from the transcript. ex "pay"	VARCHAR(64
noun	This is the noun extracted from the transcript, ex: "bill"	VARCHAR(64)
question	Indicates if verb/noun pair was used in a question versus a statement	BOOLEAN
speaker_name	Name of speaker if provided	VARCHAR(48)
start_ms	Milliseconds since beginning of the call this observation was detected	DECIMAL(19,0)
end_ms	Milliseconds since beginning of the call this observation ended	DECIMAL(19,0)
start_ratio	START_MS/Total length of the call	DECIMAL(5,3)
verb_negative	A boolean value to describe the context in which the verb is used. For example, TRUE for : I dont want to close my account.	VARCHAR(64)
noun_negative	A boolean value to describe the context in which the Noun is used. For example, TRUE for : No worries.	VARCHAR(64)

Category Observation

Information for when a calculated category was observed within an interaction, both matching and non-matching. Categories that do not match a turn (either one person talking in a dialogue) have no segment of the conversation allocated to themselves. It is essentially mapping categories to various segments across the entire conversation.

This table joins with Content, Segment and Ontology tables.

Table Name	Primary Key	Join Keys
Category Observation	Obvesrvation ID	Content_Id, Segment_id, Ontology_Id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
OBSERVATION_ID	Generated reproducible hash code,: MurmurHash(mediaId+“Category”+ categoryName)	DECIMAL(19,0)
OBSERVATION_TYPE	“CATEGORY”	VARCHAR(48)
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
ONTOLOGY_ID	PK of ONTOLOGY	BIGINT
SEGMENT_ID	Reference to turn segment if there is one else null	DECIMAL(19,0)
CATEGORY_MATCH	True if category value is 1, false otherwise (1 = Match, 0 = Not a Match, Null = category was not applied)	BOOLEAN
SPEAKER_NAME	Name of speaker if provided	VARCHAR(48)
start_ms	Milliseconds since beginning of the call this observation was detected	DECIMAL(19,0)
end_ms	Milliseconds since beginning of the call this observation ended	DECIMAL(19,0)
start_ratio	A mesure of the amount of time elapsed from the start of the conversation that the observation was detected	DECIMAL(5,3)
START_RATIO	START_MS/Total length of the call	DECIMAL(5,3)

Entity Observation

A mapping of items-- Names, places, people, things, etc extracted from the voice or messaging interaction by our classifier engine.

TABLE_NAME	PRIMARY KEY	JOIN KEYS
Entity Observation	Observation_id	Content_Id, Ontology_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Uniqie ID per voice or messaging conversation for VoiceBase engine for the observation	DECIMAL(19,0)
observation_type	A marker for the type of observation. In this case, "Entity"	VARCHAR(48)
content_id	Uniqie ID per voice or messaging conversation for VoiceBase engine for the content or media	DECIMAL(19,0)
ontology_id	Primary Key and reference to the Ontology table	BIGINT
segment_id	Reference to segment if there is one	DECIMAL(19,0)
entity_type	Type of entity, e.g. "person", "organization", ...	VARCHAR(24)
speaker_name	Name of speaker if provided	VARCHAR(48)
start_ms	Milliseconds since beginning of the call this observation was detected	DECIMAL(19,0)
end_ms	Milliseconds since beginning of the call this observation ended	DECIMAL(19,0)
start_ratio	START_MS/Total length of the call	DECIMAL(5,3)
text_value	Actual text of the entity	VARCHAR(128)
formatted_text_value	Formatted version of the entity	VARCHAR(255)
person_first_name	If type "person", the detected first name	VARCHAR(64)
person_middle_name	If type "person", the detected middle name	VARCHAR(64)
person_last_name	If type "person", the detected last name	VARCHAR(64)
money_currency	If type "currency", the detected currency	VARCHAR(64)
money_major_unit	If type "currency", the detected major unit, e.g. for $12.49 this would be 12	DECIMAL(19,0)
money_minor_unit	If type "currency", the detected minor unit, e.g. for $12.49 this would be 49	DECIMAL(5,0)

Keyword Observation

A listing of keywords extracted by our classier models. We recommend that you use Verb-Noun observations for a more nuanced and insight-rich analytics experience. For general keyword discovery intents, this table works fine.

TABLE_NAME	PRIMARY KEY	JOIN KEYS
Keyword Obersavation	Observation_id	Content_Id, Segment_id, Topic_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Uniqie ID per voice or messaging conversation for VoiceBase engine for the observation	DECIMAL(19,0)
observation_type	A marker for the type of observation. In this case, "Keyword"	VARCHAR(48)
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
segment_id	Reference to segment if there is one	DECIMAL(19,0)
name	Actual Extracted keyword	VARCHAR(128)
score	Only mandatory if we always calculate score, need to clarify Valid value range: [0,1]	DECIMAL(10,3)
topic_id	OBSERVATION_ID of topic this (topic) keyword contributed to	DECIMAL(19,0)
speaker_name	Name of the speaker if provided. Either Agent or Consumer.	VARCHAR(48)
start_ms	Milliseconds since beginning of the call this observation was detected	DECIMAL(19,0)
end_ms	Milliseconds since beginning of the call this observation ended	DECIMAL(19,0)
start_ratio	At what proportion of the call was this keyword first observed. Caclulated by : START_MS/Total length of the call	DECIMAL(5,3)

Metric Observation

A listing of various conversation metrics and their values captured against the voice or messaging interaction.

TABLE_NAME	PRIMARY KEY	JOIN KEYS
Metric Observation	Content_id	Content_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Generated reproducible hash code,: MurmurHash(mediaId+“DETECTOR”+ segmentSequence)	DECIMAL(19,0)
observation_type	“METRIC”	VARCHAR(48)
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
ontology_id	PK of ONTOLOGY	BIGINT
metric_value	Actual numeric value of the specified metric	DECIMAL(21, 3)

Sentiment Observation

A segment-wise breakup of the voice or messaging interaction and calculated sentiment score for those turn-level sections. This is expressed as any decimal value between -1 and 1. Where '-1' is bad and '+1' is good.

TABLE_NAME	PRIMARY KEY	JOIN KEYS
Sentiment Observation	Content_Id, Segment_id	Content_Id, Segment_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Generated reproducible hash code,: MurmurHash(mediaId+“SENTIMENT”+ segmentSequence)	DECIMAL(19,0)
observation_type	“SENTIMENT”	VARCHAR(48)
content_id	Uniqie ID per voice or messaging conversation for VoiceBase engine for the content or media	DECIMAL(19,0)
segment_id	Reference to segment if there is one	DECIMAL(19,0)
sentiment_value	Expressed as any decimal value between -1 and 1. Where '-1' is bad and '+1' is good.	DECIMAL(4, 3)
speaker_name	Name of speaker if provided	VARCHAR(48)
start_ms	Milliseconds since beginning of the call this observation was detected	DECIMAL(19,0)
end_ms	Milliseconds since beginning of the call this observation ended	DECIMAL(19,0)
start_ratio	At what proportion of the call was this feature first observed. Caclulated by : START_MS/Total length of the call	DECIMAL(5,3)

Topic Observation

A table that details the general topic bucket that a voice or messaging conversation may fall into. These are built based on the Keywords extracted from the interaction. As is the case with the keywords table, we recommend using the VerbNoun Observation table for a richer analytics experience. However, topics are still great for a general analytics overview of the conversation.

TABLE_NAME	PRIMARY KEY	JOIN KEYS
Topic Observation	observation_id	Content_Id, Parent_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Uniqie ID per voice or messaging conversation for VoiceBase engine for the observation	DECIMAL(19,0)
observation_type	A marker for the type of observation. In this case, "Topic"	VARCHAR(48)
content_id	Uniqie ID per call for Analytics Studio platfrom based on the media and found in the Content table	DECIMAL(19,0)
name	Name of the topic. These are often single words or a long text string formed by concatanating a group of keywords.	VARCHAR(128)
parent_id	OBSERVATION_ID of parent topic, if entry is a sub-topic. Mandatory for sub-topics as this is the only indicator if a topic is indeed a sub-topic.	DECIMAL(19,0)

Conversation Survey

Information about all surveys that have taken place.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Conversation Survey	Content_Id	Survey_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
survey_id	Uses a hash (MurrmurHash), possibly of mediaId+seq_num	BIGINT
survey_type	Indicates the survey types e.g. "Satisfaction", "PostSurvey"	VARCHAR(40)
survey_status	Indicates the survey status e.g. "completed", "skipped", "closed by consumer"	VARCHAR(40)
question	Indicates the questions asked in the during survey e.g. "How would you rate your overall satisfaction with the service you received?"	VARCHAR(255)
answer	Indicates the answer given in survey e.g. "skip","5"	VARCHAR(1000)
question_id	Indicates the question id of the questions asked in survey	VARCHAR(40)
question_type	Indicates the question type e.g. "fcr"	VARCHAR(40)
valid_answer	Indicates is answer is valid or not	BOOLEAN

Deleted Content

Information about media that was deleted from the system

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Deleted Content	Content_id	Content_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
media_id	Unique ID of of the deleted media file	VARCHAR

Engagement Agent Participants

Information about all the agent related to engagements

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Engagement Agent Participants	Content_Id,Agent_Id	Agent_participant_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
agent_participant_id	Generated reproducible hash code,: MurmurHash(mediaId +“/”+ agentId + "/" + agentRole)	BIGINT
agent_id	Specify agent Id	VARCHAR(64)
agent_full_name	Specify agent Full Name	VARCHAR(64)
agent_group_name	Specify agent Group Name	VARCHAR(64)
agent_group_id	Specify agent Group Id	BIGINT
agent_name	Specify agent Nick Name	VARCHAR(64)
agent_login_name	Specify agent Login Name	VARCHAR(64)
agent_type_name	Specify Agent User Type Name e.g. :0 "Human" or "System", "Bot"	VARCHAR(10)
agent_role	Specify agent role, e.g. "AGENT", "AGENTMANAGER"	VARCHAR(40)
agent_responses	Count of responses from the agent in wihtin the interation	DECIMAL(4,0)
agent_permission	Permissions granted to the participating agents	VARCHAR(24)
seq_num	Order in which agents joined the conversation. This should be 1 for the first agent who joined and increment by 1.	DECIMAL(2,0)
is_first_human	TRUE for the first human agent to join the conversation, FALSE for all other agents in the conversation	BOOLEAN
is_last_human	TRUE for the last human agent to join the conversation, FALSE for all other agents in the conversation	BOOLEAN
is_first_bot	TRUE for the first bot agent who join the conversation, FALSE for all other agents in the conversation	BOOLEAN
is_last_bot	TRUE for the last bot agent who join the conversation, FALSE for all other agents in the conversation	BOOLEAN
is_first	TRUE for the first human or bot agent to join the conversation, FALSE for all other agents in the conversation	BOOLEAN
is_last	TRUE for the last human or bot agent to join the conversation, FALSE for all other agents in the conversation	BOOLEAN
first_start_ms	Time elapsed since the start of the engagement in milliseconds.	DECIMAL(19,0)

Engagement Campaign Info

Information about all the campaign of the engagements.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Engagement Campaign Info	Campaign_Id,Content_Id	Content_Id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Unique id of media file	DECIMAL(19,0)
campaign_id	ID of the campaign from MIA response	VARCHAR(64)
campaign_engagement_id	ID of the campaign’s engagement from MIA response	VARCHAR(64)
campaign_engagement_name	Name of the campaignEngagement	VARCHAR(64)
campaign_name	Name of the campaign.	VARCHAR(64)
campaign_engagement_application_name	Name of Campaign Engagement’s application.	VARCHAR(64)
line_of_business	Information about line of business. This is a custom field.	VARCHAR(64)

Engagement Customer Info

Information about all the customers that have engaged.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Engagement Customer Info	Content_Id,Social_Id,Customer_Id	Customer_info_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Unique id of media file	DECIMAL(19,0)
customer_info_id	Generated reproducible hash code,: MurmurHash(mediaId +“/”+ customerId)	BIGINT
customer_id	CustomerId specified in MIA response	VARCHAR(128)
company_size	Size of company specified in MIA response	INTEGER
customer_type	Type of custmer e.g. :- vip	VARCHAR(128)
store_number	Store # from which customer did purchasing	VARCHAR(40)
account_name	Name of the customer account	VARCHAR(128)
role	Role of customer	VARCHAR(40)
social_id	Social id of customer	VARCHAR(40)
customer_status	Status of the company e.g. cancelled	VARCHAR(256)

Engagement Info

Information about all the Conversations that have happened

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Engagement Info	Content_Id, Visitor_id, Source_id,Conversation_Id	Content_Id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
conversation_id	ID of conversation	VARCHAR(40)
close_reason	Reason for closing the conversation — by agent / consumer	VARCHAR(64)
close_reason_description	Additional information regarding the conversation close reason	VARCHAR(255)
csat_rate	CSAT Score assigned to the conversation.	DECIMAL(4,1)
device	Type of device from which the conversation was initially opened.	VARCHAR(40)
duration	Time from when the consumer started the conversation until it ended	BIGINT
end_time	The dialog end time, readable format	VARCHAR(64)
first_conversation	Whether it is the consumer’s first conversation.	BOOLEAN
is_partial	Indicates whether the conversation’s data is partial	BOOLEAN
latest_skill_id	An array of latest skill IDs, represented as numbers. The latest skill ID is the latest skill which the conversation was assigned under	DECIMAL(19,0)
latest_skill_name	Most recent skill name that the conversation was assigned to	VARCHAR(255)
mcs	Meaningful Conversation Score of the conversation.	INTEGER
operating_system	Information about the operating system of the device	VARCHAR(32)
start_time	The dialog start time, readable format.	VARCHAR(64)
full_dialog_status	Information about the status of Full Dialog	VARCHAR(64)
app_id	The name of the application	VARCHAR(128)
language	Language of the conversation’s context	VARCHAR(32)
interaction_context_id	Session ID in LivePerson system	VARCHAR(40)
time_zone	Visitor’s time zone	VARCHAR(32)
integration	The integration type: mobile-sdk, web-sdk, brand-sdk	VARCHAR(255)
integration_version	The version of the integration	VARCHAR(16)
operating_system_version	The operating system version, and distribution type (if relevant)	VARCHAR(16)
browser_version	Detailed version info of the user agent (browser or host-application)	VARCHAR(128)
visitor_id	Information about the Visitor ID	VARCHAR(40)
app_name	Information about the Engagement’s application name.	VARCHAR(64)
conversation_start_page	The page’s URL from which the conversation started	VARCHAR(2000)
conversation_start_page_title	The page’s title from which the conversation started	VARCHAR(256)
country	Name of country	VARCHAR(40)
has_purchase	Flag if purchase has done	BOOLEAN
source_id	Information about the source id	VARCHAR(40)
reason_transfer_first	The reason property gives you insight into why the conversation was transferred.	VARCHAR(255)
source_agent_full_name_transfer_first	Information about the Source Agent Full Name of First Transfer	VARCHAR(40)
source_agent_id_transfer_first	Information about the Source Agent ID of First Transfer	VARCHAR(40)
source_agent_login_name_transfer_first	Information about the Source Agent Login Name of First Transfer	VARCHAR(40)
source_agent_nick_name_transfer_first	Information about the Source Agent Nick Name of First Transfer	VARCHAR(40)
source_skill_id_transfer_first	Information about the Source Skill ID of First Transfer	VARCHAR(40)
source_skill_name_transfer_first	Information about the Source Skill Name of First Transfer	VARCHAR(40)
target_skill_id_transfer_first	Information about the Target Skill ID of First Transfer	VARCHAR(40)
target_skill_name_transfer_first	Information about the Target Skill Name of First Transfer	VARCHAR(40)
transfer_time_first	Information about the First Transfer TIme	VARCHAR(40)
reason_transfer_last	Information about the Reason of last Transfer	VARCHAR(255)
source_agent_full_name_transfer_last	Information about the Source Agent Full Name of last Transfer	VARCHAR(40)
source_agent_id_transfer_last	Information about the Source Agent ID of Last Transfer	VARCHAR(40)
source_agent_login_name_transfer_last	Information about the Source Agent Login Name of lastTransfer	VARCHAR(40)
source_agent_nick_name_transfer_last	Information about the Source Agent Nick Name of last Transfer	VARCHAR(40)
source_skill_id_transfer_last	Information about the Source Skill ID of last Transfer	VARCHAR(40)
source_skill_name_transfer_last	Information about the Source Skill Name of lastTransfer	VARCHAR(40)
target_skill_id_transfer_last	Information about the Target Skill ID of Last Transfer	VARCHAR(40)
target_skill_name_transfer_last	Information about the Target Skill Name of last Transfer	VARCHAR(40)
transfer_time_last	Information about the last Transfer TIme	VARCHAR(40)
total_responses	Max amount of conversations in the response	INTEGER
latest_agent_name	The agent’s login name	VARCHAR(64)
latest_agent_id	Most recent agent ID the conversation was assigned to.	VARCHAR(64)
latest_agent_group	Group name of the agent most recently assigned to the conversation	VARCHAR(64)
latest_consumer_participant_id	Contains information about the latest consumer(s) participating id in the conversation	VARCHAR(64)
nps_text	NPS_Text assigned to the conversation	VARCHAR(64)
nps	NPS assigned to the conversation	DECIMAL(2,0)
fcr_text	FCR_Text assigned to the conversation.	VARCHAR(64)
fcr	Values of FCR (First Call Resolution) assigned to the conversation.	DECIMAL(2,0)
csat_text	CSAT_Text value of the conversation	VARCHAR(64)
csat	CSAT score of the conversation	DECIMAL(2,0)
transfer_count	Information about the transfer count	DECIMAL(4,0)
first_intent_name	The id of the intent that was created based on the first intent-able message in the conversation	VARCHAR(255)
first_intent_label	The label of the intent that was created based on first intent-able message in the conversation	VARCHAR(255)
conversation_status	Status of the conversation. EGL	NUMERIC(1,0)
conversation_end_time	Timestamp of when the conversation ended	TIMESTAMP
has_human_agent	A boleen value to flag if a human agent was involved in the engagement	BOOLEAN
has_bot	A boleen value to flag if a bot agent was involved in the engagement	BOOLEAN

Intent Observation

Information about something that the system observed happened during a conversation.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Intent Observation	Content_Id	Observation_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Unique id of media file	DECIMAL(19,0)
observation_id	Generated reproducible hash code,: MurmurHash(segmentSequence + mediaId +“INTENT”+ intentName)	BIGINT
observation_type	"INTENT"	VARCHAR(6)
ontology_Id	Primary Key of ONTOLOGY	BIGINT
name	Contains the intent name, generally its some uuid	VARCHAR(255)
label	Contains the intent of sentence e.g. "ask about discount or promotion"	VARCHAR(255)
confidence_score	Intent confidence level value as calculated by the integrated platform.	DECIMAL(8,3)
primary_intend	Determine whether its first intent in the list. Acertained by picking up the first possible intent the system picks up in a conversation	BOOLEAN
seq_num	Determines the sequence of the intent in the list	DECIMAL(4,0)
is_first_defined	Boolean to check if the first intent is indeed exists	BOOLEAN
meta_intent	It is a group of intents that have a similar nature. For example there can be a "check bill date" intent and a "check bill amount" intent that do two separate things, but are grouped under a meta intent called "Billing".	VARCHAR(255)
is_defined	Boolean to check if an intent is defined. The definition of defined is: the LABEL is not null and not the string "undefined" (case insensitive).	BOOLEAN
is_last_defined	Boolean. True, if it is the last defined intent in the list of intents activated against a conversation.	BOOLEAN

Skill Transfers

Information about skill based transfers of conversations

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Skill Transfers	content_id, source_skill_id	skill_transfer_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
SKILL_TRANSFER_ID	Unique ID for the transfer.	DECIMAL(19,0)
content_id	Unique ID for the Conversation	DECIMAL(19,0)
REASON	What type of transfer was this, for example a Skill transfer to another skill, or an Agent transfer to another agent (for example the conversation is on the correct skill but another agent was handling it, or the current agent is stepping away)	VARCHAR(64)
SOURCE_SKILL_ID	Unique Id for the skill name that the conversation started on.	DECIMAL(19,0)
SOURCE_SKILL_NAME	The skill that the conversation was on when the transfer was initiated.	VARCHAR(64)
SOURCE_AGENT_ID	Unique ID of agent who was in control of the conversation when the transfer was initiated	VARCHAR(64)
TARGET_SKILL_ID	Unique ID of the skill to which the conversation was transfered.	DECIMAL(19,0)
TARGET_SKILL_NAME	If it is a skill to skill transfer then this will be the skill that the conversations was transferred to.	VARCHAR(64)
TARGET_AGENT_ID	Unique ID of the agent to which the conversation was transfered.	VARCHAR(64)
TRANSFER_TIME_MS	UTC Time stamp at which conversation was transfered in milliseconds Unix Epoc Time	DECIMAL(19,0)
START_MS	Milliseconds since beginning of the interaction that this transfer was made	DECIMAL(19,0)
SEQ_NUM	The number of transfers that took place in one conversation, in a sequence starting from 1.	DECIMAL(4,0)
IS_FIRST	Boolean. True if it is the first transfer in a conversation.	BOOLEAN
IS_LAST	Boolean. True if it is the last transfer in a conversation.	BOOLEAN
TRANSFER_TIME	UTC Time stamp at which conversation was transfered in seconds Unix Epoc Time	DECIMAL(19,0)

Classifier Observation

Information about classifier results. If no classifier was used there should be no entry.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Classifier Observation	Content_Id, Ontology_id	Observation_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Generated reproducible hash code. Hash(mediaId+“CLASSIFIER”+ segmentSequence)	DECIMAL(19,0)
observation_type	Type of observation (e.g. 'CATEGORY', 'DETECTOR', “CLASSIFIER”...). In this case, it would be “CLASSIFIER”	VARCHAR(48)
content_id	Uniqie ID per call	DECIMAL(19,0)
ontology_id	PK of ONTOLOGY	BIGINT
class_name	Custom Name given to classifer model	DECIMAL(10,0),
class_label	This is the index ID of the calculated classification	VARCHAR(128)
score	A confidence score against the class assinged. Valid value range [-99.999,+99.999]	DECIMAL(10,3)
version	Classifier version number	VARCHAR(64),

Engagement Attachments

Information about files and other attachments uploaded or linked in the interaction.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Engagement Attachments	content_id, participant_id	engagement_attachment_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
engagement_attachment_id	Unique id of media file. Also PK for this table	DECIMAL(19,0)
content_id	Unique id of media file	DECIMAL(19,0)
type	"Link" or "File" depending on what's uploaded.	VARCHAR(16)
file_type	File type of the file uploaded	VARCHAR(96)
caption	Any caption text added aloing with the file upload	VARCHAR(255)
location	Link or path of the uploaded file	VARCHAR(1024)
speaker_name	Name of speaker if provided	VARCHAR(64)
speaker_role	Role of speaker if provided	VARCHAR(64)
participant_id	Generated reproducible hash code,: MurmurHash(mediaId +“/”+ agentId + "/" + agentRole)	VARCHAR(64)

Engagement Purchase

Information about purchases that took place as part of the interaction.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Engagement Purchase	content_id, order_id	purchase_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
purchase_id	Generated reproducible hash code, suggested: MurmurHash(mediaId +“/”+ orderId)	BIGINT
order_id	Extracted from response, Contains order id for each order	VARCHAR(40)
quantity	Value is always 1	INTEGER
total	contais total order value e.g. 78.0	DECIMAL(10, 2)
currency	If type "currency", the detected currency value.	VARCHAR(16)
cart_total	Total value of the customers cart	DECIMAL(10, 2)
cart_currency	Currency type attached to the customer's cart	VARCHAR(16)

Engagement Purchase Item

Information about the purchase items themselves in conjunction to the purchase data.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Engagement Purchase Item	content_id, purchase_id	purchase_item_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
purchase_item_id	Unique ID assigned to item purchased	DECIMAL(19,0)
purchase_id	Unique ID for the purchase itself	BIGINT
product_name	Name of product purchased if provided	VARCHAR(255)
product_sku	SKU of product if provided	VARCHAR(40)
product_category	Open text for category of product if provided	VARCHAR(255)
product_price	Numeric amount for price of product if provded	DECIMAL(10,2)
quantity	Quantity of product if provided.	DECIMAL(6,0)

Detector Observation

Information about detectors that fired and their scores. If no detector fired there should be no entry and hence no parquet file either.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Detector Observation	Content_Id, Segment_id, Ontology_Id	observation_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
observation_id	Generated reproducible hash code, suggested: MurmurHash(mediaId+“DETECTOR”+ segmentSequence)	DECIMAL(19,0)
observation_type	Type of observation (e.g. 'CATEGORY', 'DETECTOR', “CLASSIFIER”...). In this case, it would be “DETECTOR”	VARCHAR(48)
content_id	Uniqie ID per call for EA product	DECIMAL(19,0)
ontology_id	PK of ONTOLOGY	BIGINT
segment_id	Reference to segment if there is one	DECIMAL(19,0)
class_name	Index number of the classifier	DECIMAL(10,0)
class_label	Lablel of the classifer if available. EG: "PCI"	VARCHAR(128)
version	Version number of classifier used	VARCHAR(64)
speaker_name	Name of speaker if provided	VARCHAR(48)
start_ms	Milliseconds since beginning of the call this observation was detected	DECIMAL(19,0)
end_ms	Milliseconds since beginning of the call this observation ended	DECIMAL(19,0)
start_ratio	START_MS/Total length of the call	DECIMAL(5,3)

Agent Surveys

Agent Survey as filled in by the agent either during or after the conversation has ended, and lets the agent record things like the nature or outcome of the conversation.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Agent Surveys	survey_id, ac_survey_id	agent_survey_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
agent_survey_id	Unique ID for the survey record	DECIMAL(19,0)
content_id	Unique id for the record across platform	DECIMAL(19,0)
survey_status	Operational status of the survey. EG: Open, Closed	VARCHAR(40)
status_reason	Cause for the provided status. EG: If the survey_status is "CLOSED" then the status_reson might be "TIMEOUT". Ie. the survey ended due to inactivity or a set timeout.	VARCHAR(64)
survey_id	Unique ID per survey	VARCHAR(40)
ac_survey_id	A unique ID for the survey per LivePerson Platform	VARCHAR(40)
ac_survey_name	A select name for the survey per LivePerson Platform	VARCHAR(40)
ac_survey_revision	A unique ID for the survey per LivePerson Platform	DECIMAL(4,0)
survey_skill_id	Skill ID assigned to the survey	DECIMAL(19,0)
survey_skill_name	Skill Name assigned to the survey	VARCHAR(64)
assigned_agent_id	ID of the agent assigned to the survey	VARCHAR(64)
performed_agent_id		VARCHAR(64)
last_update_time	Timestamp of when the last update (Unix Epoc Seconds)	DECIMAL(19,0)
is_latest_survey	Boolean to check if its the latest survey	BOOLEAN

Agent Survey Questions

Information about the questions marked against a survey.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Agent Survey Questions	content_id, agent_survey_id	agent_survey_question_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
agent_survey_question_id	Unique ID per survey question	DECIMAL(19,0)
content_id	Uniique ID for VB Plaftorm	DECIMAL(19,0)
agent_survey_id	Unique ID of the survey to which the question is attached	DECIMAL(19,0)
question_text	Survey question text	VARCHAR(255)
question_id	Unique ID per question	VARCHAR(40)
question_definition	Defiition/ Description of the question	VARCHAR(40)
question_category	Open text category to which the survey belongs	VARCHAR(40)
is_conversation_outcome	if TRUE, said question marks the outcome of the survey.	BOOLEAN

Agent Survey Answers

Information about the answer provided within a survey.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Agent Survey Answers	content_id, agent_survey_id	agent_survey_answer_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
agent_survey_answer_id	Unique ID per survey answer	DECIMAL(19,0)
content_id	Uniique ID for VB Plaftorm	DECIMAL(19,0)
agent_survey_id	Unique ID of survey to which answer is attached	DECIMAL(19,0)
answer	Survey answer text	VARCHAR(255)
answer_id	Unique ID per answer	VARCHAR(40)

Agent Status Event

Information about change in agent's status ie. Online, offline, away and such.

TABLE_NAME	JOIN KEYS	PRIMARY KEY
Agent Status Event	agent_status_event_id, agent_id	agent_status_event_id

FIELD NAME	DESCRIPTION	DATA TYPE & LIMIT
agent_status_event_id	Unique ID per event	DECIMAL(19,0)
account_id	Unique ID per VB account	VARCHAR(36)
agent_id	Unique ID per agent in alpha numeric format	VARCHAR(64)
agent_id_numeric	Unique ID per agent in strictly numeric format	DECIMAL(19,0)
status_change_time	Timestamp of status update	TIMESTAMP
session_id	Unique ID for a given conversation between customer and agent. ie. Identifier of the session during which this status change took place	DECIMAL(19,0)
sequence_number	Sequence number that prescribes the order of each message / event	INTEGER
status_type	Type of status change. 1 - status changed, see `statusSubType` 3 - login 4 - logout	SMALLINT
status_sub_type	Subtype of status change with statusType=1 . 1 - offline 2 - online 3 - occupied 4 - away	SMALLINT
status_reason_id	Identifier of optional custom reason for the status change	DECIMAL(19,0)
status_reason_text	Optional custom reason for the status change	VARCHAR(64)
prev_status_change_time	Time of this agent’s previous status change	TIMESTAMP