version: 2 models: - name: clip_engagement_features description: "Mart model that aggregates all engagement signals (plays, likes, shares, pins, etc.) into a comprehensive feature set for each clip. This model serves as a foundational layer for downstream recommendation and analytics tasks." config: group: recommendations tags: ["cadence=daily", "dbt"] columns: - name: clip_id description: "Unique identifier for the clip" - name: clip_creator_user_id description: "ID of the user who created the clip" - name: first_play_start_ts_utc description: "Timestamp of the very first play of this clip by any user" - name: last_play_start_ts_utc description: "Timestamp of the most recent play of this clip by any user" - name: last_engagement_timestamp description: "The latest timestamp from any engagement event (play, like, share, pin, etc.) for this clip. Used for incremental processing." - name: n_creator_likes description: "Number of likes the creator has given their own clip" - name: n_creator_verified_likes description: "Number of likes from the creator where they listened for at least 5 seconds" - name: creator_net_pin_clip description: "Net pin action from the creator on their own clip (1 for pinned, 0 for not)" - name: n_creator_shares description: "Number of times the creator has shared their own clip" - name: did_owner_add_to_playlist description: "Boolean indicating if the creator has added their own clip to a playlist" - name: n_playlists_added_by_owner description: "Number of unique playlists the creator has added their clip to" - name: n_non_creator_likes description: "Number of likes from users other than the creator" - name: n_non_creator_verified_likes description: "Number of likes from non-creators where they listened for at least 5 seconds" - name: n_non_creator_shares description: "Number of shares from users other than the creator" - name: n_unique_non_creator_users_shared description: "Number of unique non-creator users who have shared this clip" - name: n_non_creator_added_to_playlist description: "Number of non-creator users who have added this clip to a playlist" - name: n_playlists_added_by_non_owner description: "Number of unique playlists this clip has been added to by non-creators" - name: public_clip_static_features description: "Mart model for public clip static features used in recommendation systems" config: group: recommendations tags: ["cadence=daily", "dbt"] columns: # Base clip data from dim_clip - name: clip_id description: "Unique identifier for the clip" - name: p_date description: "Partition date for the clip" - name: p_hour description: "Partition hour for the clip" - name: created_at description: "Timestamp when the clip was created" - name: user_id description: "ID of the user who created the clip" - name: request_id description: "Unique request identifier" - name: s3_id description: "S3 identifier for the clip file" - name: platform description: "Platform where the clip was created" - name: display_index description: "Display order index for the clip" - name: initial_title description: "Initial title of the clip" - name: time_used_sec description: "Time used in seconds for clip generation" - name: type description: "Type of clip generation" - name: task description: "Specific task performed for the clip" - name: priority description: "Priority level of the clip request" - name: duration_sec description: "Duration of the clip in seconds" - name: is_refund_credits description: "Whether credits were refunded for this clip" - name: is_bot description: "Whether the clip was created by a bot" - name: credit_cost description: "Number of credits used for this clip" - name: lyrics_at_gen description: "Lyrics at generation time" - name: is_simple description: "Whether this is a simple clip" - name: simple_prompt description: "Simple prompt used for generation" - name: lyrics_final description: "Final lyrics of the clip" - name: is_make_instrumental description: "Whether the clip is instrumental" - name: style_tags description: "Style tags associated with the clip" - name: negative_style_tags description: "Negative style tags for the clip" - name: request_models description: "Models requested for generation" - name: model_name description: "Name of the model used for generation" - name: experiment_inf_param description: "Experiment inference parameters" - name: experiment_version description: "Version of the experiment" - name: history description: "JSON history of the clip generation process" - name: concat_history description: "Concatenated history information" - name: error_type description: "Type of error if any occurred" - name: error_message description: "Error message if any occurred" - name: upload_has_vocal description: "Whether the uploaded file has vocals" - name: edit_session_id description: "Session ID for editing" - name: edited_clip_id description: "ID of the edited clip" - name: task_details description: "Detailed task information" - name: creation_source description: "Source of clip creation" - name: augmented_tags description: "Augmented tags for the clip" - name: free_quota_category description: "Category for free quota usage" - name: lyrics_language description: "Language of the lyrics" - name: simple_lyrics_model description: "Model used for simple lyrics" - name: created_session_id description: "Session ID when clip was created" - name: is_audio_watermark description: "Whether audio watermark is applied" - name: user_tier description: "Tier of the user" - name: is_deleted description: "Whether the clip is deleted" # Additional columns from dim_clip - name: clip_uid description: "Unique identifier for the clip" - name: cleaned_create_mode description: "Cleaned creation mode - derived from create_mode or is_simple flag" - name: model_category description: "Categorized model version (v3.5, v4, v4.5, stem, etc.)" - name: category description: "Categorized clip type (normal_gen, cover_gen, extend_gen, etc.)" - name: history_length description: "Number of items in the history JSON array" - name: is_made_in_edit description: "Whether the clip was made in edit mode" - name: has_augmented_tags description: "Whether the clip has augmented tags" - name: was_prompt_used_exactly description: "Whether the prompt was used exactly" - name: owner_creation_day_total_clip_cnt description: "Total number of clips created by the owner on their creation day" - name: owner_creation_day_play_duration_sec_own description: "Total play duration in seconds for the owner on their creation day" - name: owner_creation_day_play_duration_sec_others description: "Total play duration in seconds for others on the owner's creation day" - name: owner_creation_day_subscription_tier description: "Subscription tier of the owner on their creation day" - name: like_count description: "Number of likes for the clip" - name: dislike_count description: "Number of dislikes for the clip" - name: flag_count description: "Number of flags for the clip" - name: play_count_displayed description: "Number of times the clip was played" - name: download_audio_mp3_count description: "Number of times the clip was downloaded as MP3" - name: download_audio_wav_count description: "Number of times the clip was downloaded as WAV" - name: download_video_count description: "Number of times the clip was downloaded as video" - name: share_count description: "Number of times the clip was shared" - name: owner_init_play_cnt description: "Number of plays for the owner" - name: owner_init_play_duration_seconds description: "Total play duration in seconds for the owner" - name: owner_init_net_like_events description: "Net like events for the owner (likes minus unlikes)" - name: owner_init_net_dislike_events description: "Net dislike events for the owner (dislikes minus undislikes)" - name: owner_init_n_share_events description: "Number of share events for the owner" - name: owner_init_n_download_events description: "Number of download events for the owner" - name: owner_init_net_public_toggle_events description: "Net public toggle events for the owner (make public minus make private)" - name: owner_init_net_positive_interactions description: "Net positive interactions for the owner" - name: owner_init_satisfaction_metric_version description: "Version of the satisfaction metric used" - name: owner_init_is_satisfactory_indirect description: "Whether the clip is satisfactory based on indirect metrics" - name: owner_init_is_satisfactory_direct description: "Whether the clip is satisfactory based on direct metrics" - name: owner_init_is_satisfactory description: "Whether the clip is satisfactory overall" - name: first_play_start_ts_utc description: "Timestamp of the first play" - name: last_play_start_ts_utc description: "Timestamp of the last play" - name: creator_first_play_start_ts_utc description: "Timestamp of the first play from the creator" - name: creator_last_play_start_ts_utc description: "Timestamp of the last play from the creator" - name: creator_play_cnt_0s description: "Number of plays from the creator" - name: creator_play_cnt_5s description: "Number of plays from the creator that lasted at least 5 seconds" - name: creator_play_cnt_25pct description: "Number of plays from the creator that covered at least 25% unique seconds of the clip" - name: creator_play_cnt_50pct description: "Number of plays from the creator that covered at least 50% unique seconds of the clip" - name: creator_play_cnt_90pct description: "Number of plays from the creator that covered at least 90% unique seconds of the clip" - name: creator_play_duration_seconds_0s description: "Total play duration in seconds for the creator" - name: creator_play_duration_seconds_5s description: "Total play duration in seconds for the creator that lasted at least 5 seconds" - name: creator_play_duration_seconds_25pct description: "Total play duration in seconds for the creator that covered at least 25% unique seconds of the clip" - name: creator_play_duration_seconds_50pct description: "Total play duration in seconds for the creator that covered at least 50% unique seconds of the clip" - name: creator_play_duration_seconds_90pct description: "Total play duration in seconds for the creator that covered at least 90% unique seconds of the clip" - name: non_creator_first_play_start_ts_utc description: "Timestamp of the first play from non-creators" - name: non_creator_last_play_start_ts_utc description: "Timestamp of the last play from non-creators" - name: n_non_creator_listeners_0s description: "Number of listeners for non-creators" - name: n_non_creator_listeners_5s description: "Number of listeners for non-creators that lasted at least 5 seconds" - name: n_non_creator_listeners_25pct description: "Number of listeners for non-creators that covered at least 25% unique seconds of the clip" - name: n_non_creator_listeners_50pct description: "Number of listeners for non-creators that covered at least 50% unique seconds of the clip" - name: n_non_creator_listeners_90pct description: "Number of listeners for non-creators that covered at least 90% unique seconds of the clip" - name: non_creator_play_cnt_0s description: "Number of plays from non-creators" - name: non_creator_play_cnt_5s description: "Number of plays from non-creators that lasted at least 5 seconds" - name: non_creator_play_cnt_25pct description: "Number of plays from non-creators that covered at least 25% unique seconds of the clip" - name: non_creator_play_cnt_50pct description: "Number of plays from non-creators that covered at least 50% unique seconds of the clip" - name: non_creator_play_cnt_90pct description: "Number of plays from non-creators that covered at least 90% unique seconds of the clip" - name: non_creator_play_duration_seconds_0s description: "Total play duration in seconds for non-creators" - name: non_creator_play_duration_seconds_5s description: "Total play duration in seconds for non-creators that lasted at least 5 seconds" - name: non_creator_play_duration_seconds_25pct description: "Total play duration in seconds for non-creators that covered at least 25% unique seconds of the clip" - name: non_creator_play_duration_seconds_50pct description: "Total play duration in seconds for non-creators that covered at least 50% unique seconds of the clip" - name: non_creator_play_duration_seconds_90pct description: "Total play duration in seconds for non-creators that covered at least 90% unique seconds of the clip" - name: n_creator_likes description: "Number of likes from the creator" - name: n_creator_verified_likes description: "Number of verified likes from the creator" - name: creator_has_undo_like description: "Whether the creator has undone a like at any point" - name: n_non_creator_likes description: "Number of likes from non-creators" - name: n_non_creator_verified_likes description: "Number of verified likes from non-creators" - name: n_comments description: "Total number of non-deleted comments for the clip" - name: n_non_reply_comments description: "Number of non-deleted top-level comments (where parent_id is null)" - name: creator_trustworthiness_features description: "Mart model for creator trustworthiness features used in recommendation systems" config: group: recommendations tags: ["cadence=daily", "dbt"] columns: - name: user_id description: "ID of the user" - name: user_uid description: "Unique identifier for the user" - name: n_days_active description: "Number of days the user has been active" - name: n_days_active_create description: "Number of days the user has been active and created" - name: n_days_active_listen description: "Number of days the user has been active and listened" - name: n_days_active_l7d description: "Number of days the user has been active in the last 7 days" - name: n_days_active_create_l7d description: "Number of days the user has been active and created in the last 7 days" - name: n_days_active_listen_l7d description: "Number of days the user has been active and listened in the last 7 days" - name: n_days_active_listen_own_l7d description: "Number of days the user has been active and listened to their own clips in the last 7 days" - name: n_days_active_listen_others_l7d description: "Number of days the user has been active and listened to other users' clips in the last 7 days" - name: n_days_active_l28d description: "Number of days the user has been active in the last 28 days" - name: n_days_active_create_l28d description: "Number of days the user has been active and created in the last 28 days" - name: n_days_active_listen_l28d description: "Number of days the user has been active and listened in the last 28 days" - name: n_days_active_listen_own_l28d description: "Number of days the user has been active and listened to their own clips in the last 28 days" - name: n_days_active_listen_others_l28d description: "Number of days the user has been active and listened to other users' clips in the last 28 days" - name: n_clips_created description: "Number of clips the user has created" - name: n_gen_clips_created description: "Number of clips the user has created using the generative AI" - name: pct_clips_created_gen description: "Percentage of clips the user has created using the generative AI" - name: avg_clips_created_per_active_create_day description: "Average number of clips the user has created per day they have been active and created" - name: n_plays description: "Number of clips the user has listened to" - name: n_plays_own description: "Number of clips the user has listened to their own clips" - name: n_plays_others description: "Number of clips the user has listened to other users' clips" - name: pct_plays_own description: "Percentage of total play count the user has listened to their own clips" - name: ratio_plays_others_to_own description: "Ratio of total play count the user has listened to other users' clips to their own clips" - name: avg_plays_per_active_listen_day description: "Average number of clips the user has listened to per day they have been active and listened to clips" - name: avg_plays_own_per_active_listen_day description: "Average number of clips the user has listened to their own clips per day they have been active and listened to clips" - name: avg_plays_others_per_active_listen_day description: "Average number of clips the user has listened to other users' clips per day they have been active and listened to clips" - name: avg_plays_per_active_create_day description: "Average number of clips the user has created per day they have been active and created clips"