INSERT OVERWRITE TABLE internal_platform_db.ads_qq_sq_frd_recommendation_result_list_df_query_engine_138 PARTITION (imp_date = 20260608)
WITH feature_modified_v2 AS (
SELECT
 uin
, touin
, COALESCE(c2c_score, 0) AS c2c_score
, COALESCE(LOG(1 + c2c_cnt_score), 0) AS log_c2c_cnt_score
, COALESCE(socialzone_visit_score, 0) AS socialzone_visit_score
, COALESCE(socialzone_like_score, 0) AS socialzone_like_score
, COALESCE(socialzone_comment_score, 0) AS socialzone_comment_score
, COALESCE(profile_view_score, 0) AS profile_view_score
, COALESCE(profile_like_score, 0) AS profile_like_score
, COALESCE(touser_id_active_layer, 0) AS touser_id_active_layer
, COALESCE(LOG(100 - age_diff), 0) AS log_age_diff_complement
, COALESCE(LOG(1 + common_frd_num), 0) AS log_common_frd
, COALESCE(active_layer_score, 0) AS active_layer_score
, COALESCE(LOG(1 + frd_tag_num), 0) AS log_frd_tag_num
, COALESCE(is_same_city, 0) AS is_same_city
, COALESCE(is_focus_frd, 0) AS is_focus_frd
, COALESCE(comsg_score, 0) AS comsg_score
FROM
internal_platform_db.dwd_relationship_strength_features_v4_di_query_engine_138
WHERE
imp_date = 20260608
),
weighted_score_v2 AS (
SELECT
fm.uin
, fm.touin
, (0.0327 * fm.log_frd_tag_num +
0.0386 * fm.touser_id_active_layer +
0.0853 * (1 - fm.is_same_city) +
0.0564 * fm.c2c_score +
0.1895 * fm.log_c2c_cnt_score +
0.2455 * fm.socialzone_visit_score +
0.1055 * fm.socialzone_like_score +
0.1514 * fm.socialzone_comment_score +
0.1556 * fm.profile_view_score +
0.1531 * fm.profile_like_score) / 28 AS score
FROM
feature_modified_v2 fm
)
SELECT
ws.uin
, ws.touin
, 1 + (ROUND(ws.score, 4) * 10000) AS score
, ROUND(ws.score, 4) AS raw_score
FROM
weighted_score_v2 ws
WHERE
uin >= 10000
AND touin >= 10000