Technical research on real-time multimodal interaction architecture based on AI big language model for dynamic adjustment of semantic features in intelligent customer service robots
Bibliographic record
Abstract
The service efficiency of intelligent customer service robots affects the service operation efficiency of enterprises and plays an important role in maintaining customer resources.This paper applies multimodal interaction technology to intelligent customer service system, takes multimodal big language model Qwen-VL as the core, proposes a two-stage relationship multimodal relationship extraction framework based on big language model, realizes multimodal relationship extraction with the help of high-quality auxiliary knowledge, integrates dynamic semantic features and static structural features to complete the multimodal emotion polarity prediction, and constructs multimodal retrieval Q&A system to improve the performance of smart robot performance.Applying the intelligent customer service system in this paper for service practice, the conversation between the intelligent customer service robot and the customer usually ends in about 50 rounds, and the service efficiency is relatively efficient.In the face of customer emotional sentences labeled as happy, complaining and angry, the recognition accuracy under multimodal sentiment analysis is greater than 99%, and the behavior of "notification" and "confirmation" service behavior accounts for the largest proportion of behaviors, and the number of behaviors reaches 560,365 times, 365976 times, which is in line with the expected service behavior of intelligent customer service robots.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".