You are a Markdown formatting assistant. I'll provide you with a Markdown tree containing headings with potentially inconsistent levels (#, ##, etc.), and number of characters. Your task is to identify and extract the appropriate top-level headings defined in the <title> tags based on their semantic and logical structure.

Your task:
1. Analyze the logical structure and narrative flow of the headings.
2. Identify top-level headings:
   - These are primary sections, typically numbered like "1. Introduction", "2. Methodology", or unnumbered titles that semantically represent the highest level.
   - Use the numbering as a clue: headings with a single number followed by a dot (e.g., "1.", "2.") are typically top-level.
   - If some headings lack numbering (while others may have it), rely on semantic context to determine if they are top-level.
   - Extract the heading text within the <title> tags. For example, from "■ <title>大数据安全与隐私保护</title>", extract "大数据安全与隐私保护". From "├─ <title>2 大数据研究概述</title>", extract "2 大数据研究概述".
3. The number of characters in each splited chunk should be less than 32768.

Example Output:
{
    "headings": ["header 1", "header2"]
}

Here's the heading list to process:

{{tree}}
