(query, candiate, type = 'cal')
| 70 | |
| 71 | |
| 72 | def alignment(query, candiate, type = 'cal'): |
| 73 | #query和candiate分别是两个表，然后需要提出他们的column当作节点 |
| 74 | df_query = pd.read_csv('benchmark/' + query) |
| 75 | df_cand = pd.read_csv('benchmark/' + candiate) |
| 76 | df_query_sem = pd.read_csv('UsemLshTest/' + query) |
| 77 | df_cand_sem = pd.read_csv('UsemLshTest/'+ candiate) |
| 78 | G = nx.Graph() |
| 79 | query_columns = list(df_query.columns) |
| 80 | query_columns = [query+' '+l for l in query_columns] |
| 81 | cand_columns = list(df_cand.columns) |
| 82 | cand_columns = [candiate+' '+l for l in cand_columns] |
| 83 | G.add_nodes_from(query_columns, bipartite=0) |
| 84 | G.add_nodes_from(cand_columns, bipartite=1) |
| 85 | max_c = min(len(query_columns), len(cand_columns))#最大的c |
| 86 | for attr_s in query_columns: |
| 87 | for attr_t in cand_columns: |
| 88 | s_name = attr_s.split(maxsplit=1)[1] |
| 89 | t_name = attr_t.split(maxsplit=1)[1] |
| 90 | score = compute_ensemble_score(df_query[s_name], df_cand[t_name], df_query_sem[s_name], df_cand_sem[t_name]) |
| 91 | G.add_edge(attr_s, attr_t, weight=score) |
| 92 | |
| 93 | matching = {} |
| 94 | edges = sorted(G.edges(data=True), key=lambda x: x[2]['weight'], reverse=True) |
| 95 | max_c_scores = {} |
| 96 | c = 1 |
| 97 | mul = 1 |
| 98 | for u, v, d in edges: |
| 99 | if u in matching or v in matching.values(): |
| 100 | continue |
| 101 | matching[u] = [v, d['weight']] |
| 102 | mul = mul * d['weight'] |
| 103 | max_c_scores[c] = mul |
| 104 | c += 1 |
| 105 | if len(matching) == max_c: |
| 106 | break |
| 107 | if type == 'distribution': |
| 108 | return max_c_scores |
| 109 | |
| 110 | goodness_score = 0 |
| 111 | best_c = 1 |
| 112 | for c, max_c_score in max_c_scores.items(): |
| 113 | c_goodness = get_c_goodness(c, max_c_score) |
| 114 | if c_goodness >= goodness_score: |
| 115 | goodness_score = c_goodness |
| 116 | best_c = c |
| 117 | sorted_matching = sorted(matching.items(), key = lambda item : item[1][1], reverse=True) |
| 118 | |
| 119 | return sorted_matching[:best_c], goodness_score |
| 120 | |
| 121 | |
| 122 | def cal_max_c_alignment(folder_path = 'benchmark'): |
no test coverage detected