Figure 2
The figure is a large multi‑panel diagram illustrating data‑driven recognition and expert‑based determination of construction law (C L) documents for the C L K R. The top band, labeled on the right “(a) Data‑driven recognition of candidate documents for C L K R”, shows a three‑step pipeline. The first box on the left reads “Collect corpora containing C L documents from China Judgements Online”, and beneath it, within a rounded rectangle, “374,992 written judgments” with a small screenshot icon. A right‑pointing arrow leads to the second box “Recognize document name entities by guillemets (left and right double angle brackets)” with a count “775,241 document name entities”. Another arrow leads to “Cleanse the identified document name entities with three criteria” with three stacked numbers: “7,954 non‑duplicate items”, “1,018 items with no less than 5 appearances”, and “702 candidate documents that end with 10 specific terms”. A downward label “Provide 702 candidate documents” points to the lower half of the figure. The bottom half, marked on the right as “(b) Expertise‑based determination of C L documents in C L K R”, is split into three main text boxes at the top and a large visualization underneath. From left to right, the boxes state: “Filter and align the 702 candidate documents”, “Clarify the 8 C L knowledge areas and 164 C L knowledge subareas”, and “Categorize 387 C L documents into 164 distinct C L K subareas (Table S 2)”. Below the first box is a grey panel headed “387 C L documents for C L K R” containing green rounded rectangles labeled “C L D‑001”, “C L D‑002”, “C L D‑003”, “C L D‑004”, “C L D‑005”, “C L D‑006”, followed by dotted ellipsis dots and ending with “C L D‑386” and “C L D‑387”. The center of the figure depicts “C L Knowledge” as a circle feeding eight colored second‑layer areas labeled “C 1: Basic Legal Knowledge for Construction”, “C 2: Construction Permits”, “C 3: Contracting and Subcontracting”, “C 4: Construction Project Contracts and Labor Contracts”, “C 5: Environment and Cultural Heritage Protection”, “C 6: Construction Safety”, “C 7: Construction Quality”, and “C 8: Dispute Resolution”. Each C‑area connects to a band of thinner third‑layer labels such as “C 1‑01 to C 1‑29”, “C 2‑01 to C 2‑14”, “C 3‑01 to C 3‑16”, “C 4‑01 to C 4‑23”, “C 5‑01 to C 5‑12”, “C 6‑01 to C 6‑25”, “C 7‑01 to C 7‑23”, and “C 8‑01 to C 8‑22”. From these subarea codes, many multicolored strands flow rightward into a tall rectangular block labeled on the side “387 C L documents in C L K R”, whose interior is filled with vertical green rectangles representing individual documents. At the bottom, a dashed “Legends” box explains icons: open rounded rectangles represent “Second‑layer C L K area”, narrow rounded rectangles indicate “Third‑layer C L K subarea”, and solid green bars denote “Construction law document”. To the right of the legend is a worked example titled “An example of C L knowledge in subarea C 3‑06”. It shows “C L K” leading to “C 3: Contracting and Subcontracting”, then to “C 3‑06: Statutory requirements for winning bids and handling complaints in bidding”, which in turn connects to three specific green document bars labeled: “C L D‑072: Opinions on Promoting the Sustainable and Healthy Development of the Construction Industry”, “C L D‑232: Tendering and Bidding Law of the People’s Republic of China”, and “C L D‑266: Regulations for the Implementation of Bidding and Tendering Law of the People’s Republic of China”. The entire diagram emphasizes the progression from hundreds of thousands of judgments to a curated set of 387 construction law documents structured into 8 knowledge areas and 164 subareas within the C L K R.Building the CLKR by combining data-driven and expertise-based paradigm. Source(s): Authors’ own work