In the context of the Chinese civil-service Shenlun (申论) writing examination — administered nationally through the Guokao (国家公务员考试) by the State Administration of Civil Service and provincially in parallel sittings — "merge" (归并, guībìng, or 合并, hébìng) is a core processing technique within the broader summarisation family of tasks. Shenlun questions present candidates with a packet of given materials (给定资料) and demand answers built strictly from those materials. The merge operation is the second stage of the standard workflow of extract, merge, refine (提炼、归并、加工): after a candidate underlines and lifts raw points from the passages, those points are combined so that semantically identical or causally related fragments collapse into a single consolidated entry, eliminating redundancy and exposing the underlying logical structure the graders' rubric rewards.
Merging operates by three recognised principles. First, merge by identical meaning (同义合并): points that restate the same idea in different wording — common because Shenlun materials deliberately scatter paraphrases of one fact across several paragraphs — are fused into one. Second, merge by category (同类合并): points sharing an attribute are grouped under a superordinate heading, such as collapsing several specific shortfalls into "institutional causes," "economic causes," and "ideological causes." Third, merge by subject or actor (同主体合并): points are sorted by the entity responsible — government, enterprise, society, individual — which is the dominant framework for countermeasure (对策) questions. Effective merging requires the candidate to abstract a key term (关键词) or topic sentence and front-load it, because Chinese answer-scoring (采分点, points-based marking) credits the visible presence of the rubric keyword. A point buried inside prose without its category label often forfeits the mark even when the substance is present.
The technique is tested most directly in the summarise (概括/归纳) and comprehensive analysis (综合分析) question types, and indirectly in the propose countermeasures and the large essay (文章写作) sections, where a well-merged body of evidence supplies the paragraph architecture. A 2023 Guokao prefecture-level paper, for example, asked candidates to summarise the difficulties a rural revitalisation project faced; high-scoring answers merged a dozen scattered complaints into four labelled clusters — funding, talent, infrastructure, and coordination — rather than listing each grievance separately. Over-merging (losing distinct point-marks by fusing genuinely different ideas) and under-merging (redundant, padded answers that exceed the character limit) are the two failure modes examiners penalise; the discipline lies in matching the granularity of the merge to the question's point allocation and stated word cap.
For exam preparation, merge belongs squarely to the Shenlun paper rather than the Xingce (行测, administrative-aptitude) paper, and candidates should drill it against authentic past-paper materials with model answer-keys. The typical question angle is not "define merge" but a performance task: produce a summary under a strict character limit (often 200–400 characters) in which merging quality is the decisive differentiator between average and top-band scripts. Mastery of merge, paired with accurate extraction and concise refinement, is what converts comprehension of the materials into rubric-aligned marks.
Example
In the 2023 national Guokao Shenlun paper, top candidates merged a dozen scattered rural-revitalisation grievances into four labelled clusters — funding, talent, infrastructure, and coordination — to capture the rubric's keyword point-marks.
Frequently asked questions
Extracting (提炼) is lifting raw points from the given materials by underlining; merging (归并) is the subsequent step of combining identical or related points into consolidated, non-overlapping categories. Extraction gathers; merging organises and de-duplicates.