Skip to main content

Not applicable scenarios

When using our knowledge management and technical services, you should pay attention to the following technical limitations and inapplicable scenarios. Our technique mainly relies on embedding methods, which are mathematical representations that convert words, phrases, paragraphs, or documents into fixed-length vectors to capture rich semantic information. However, this approach may have limitations in certain situations:

tip

  • Non-text data processing limitations: Our technology mainly processes text data; its capabilities are limited for non-text content such as images, audio, and video.

  • Context dependence: Some embeddings may have difficulty capturing long-range context, affecting the understanding of complex background problems.

  • Language Difference Sensitivity: Subtle language differences, such as sarcasm, humor, or puns, may not be accurately understood.

  • Real-time data update requirements: If the knowledge base needs to be updated in real time, the embedding method may not be suitable because its training or update usually takes time.

  • Technical or terminology-intensive text processing: For texts containing a lot of technical terms, general embedding may not be accurate enough.

  • Specific language or dialect support: A few languages ​​or dialects may not be able to generate effective embeddings due to lack of pre-training data.

  • Ambiguous or incorrect language usage: Typos, grammatical errors, or non-standard usage in the text can lead to inaccurate understanding of the embedding.

  • Polysemy and ambiguity handling: There may be limitations in handling polysemy and ambiguity when determining word meaning requires broad context.

  • Personalized customization needs: The standard Embedding model may not adapt to the personalized customization needs of specific users.

The following situations may not be suitable for using our products​

  • PDF files primarily presented as images and full-image PDFs.
  • Video content learning: Our product does not currently support video content analysis.
  • Knowledge management of complex formulas: Subject documents containing complex formulas cannot be accurately identified.
  • Time-sensitive management: There may be limitations in multi-version policy documents or in judging policy effectiveness for a specific time period.
  • Complex Excel processing: Limited support for complex Excel files for structured data parsing and contextual understanding.

In order to solve these limitations, we may take the following measures in the future​

info

We will continue to work to overcome these limitations by updating and improving the product. Please learn more about the above unavailable scenarios before using the product, and seek other tools or contact our customer support team if necessary.

  • Combining multimodal learning techniques to optimize non-text data processing.
  • Use advanced large language models to capture more granular contextual information.
  • Embeddings are updated regularly to reflect new data and language trends.
  • Develop domain-specific embedding models.
  • Improve input data quality using text preprocessing tools. Please monitor updates regularly for the latest product capabilities and any newly discovered limitations. Your feedback is important for us to continuously improve our services. Thank you for your understanding and support!