private void 快速测试F5ToolStripMenuItem_Click(object sender, EventArgs e) { this.messageListBox.Items.Clear(); this.JWordSegmentor.FilterStopWords = this.checkBox1.Checked; this.JWordSegmentor.AutoInsertUnknownWords = true; this.JWordSegmentor.AutoStudy = true; this.JWordSegmentor.FreqFirst = true; string[] keywords = this.JWordSegmentor.Segment(this.TextFile()).ToArray(); Dictionary <string, ulong> dic = KwHelper.KeywordSortor(keywords, false); foreach (var item in dic) { this.AddText(item.Key + " | " + item.Value.ToString()); } }
/// <summary> /// Server程序启动 /// </summary> private void Start() { this.StatusString = "开始抓取"; ///读入一批URL列表 WebPage[] _CrawlUrls = Service.DbHelper.CrawlUrls; foreach (WebPage url in _CrawlUrls) { ///从URL链接获取该URL的HtmlCode Dictionary <string, ulong> keywords = HtmlHelper.SetupSingleUrl( url.Url, true, Service.JWordSegmentor, Service.DbHelper.WebRuleCollections); ///根据关键词匹配广告 long[] adIds = KwHelper.Match(keywords, _AdHelper.AdInvertedIndex); ///建立广告位(页面)与广告的相关联的关系文件 Service.FileService.CreateBillboardRelation(adIds, url.Url); } }