[br][b] (一)网络舆情与旅游者情绪[/b][br][br] 旅游者情绪在网络舆情中体现。随着互联网的普及,以微博、论坛、博客等为代表的网络社交媒体广泛流行,网络舆情逐渐成为影响人们情绪、态度行为的重要因素。[br][br] 网络舆情(Network Public Opinion),是指在互联网上流行的对社会问题不同看法的网络舆论,是社会舆论的一种表现形式,是通过互联网传播的公众对生活中某些热点、焦点问题所持的有较强影响力、倾向性的言论和观点。它具有以下几种特征。[br][br] 1. 直接性[br][br] 直接性是指网民可以通过微博、论坛和博客随时发表意见,民意表达十分畅通;网络舆论具有无限次即时快速传播的可能性,网民可以转发将信息重新传播,一个爆炸性的新闻信息能在很短的时间被大多数网民获取。[br][br] 2. 虚拟性[br][br] 互联网是一个虚拟的空间,发言者的身份是隐蔽的,再加上我国对网络舆情的管理和监督机制不够完善,因此网络舆情的真实性值得推敲。有的信息可能是网民片面、错误的认识,有的信息可能是网民宣泄情绪所捏造的,还有的信息可能是出于商业目的甚至是不法目的杜撰的。因此,网络舆情具有一定的虚拟性。[br][br] 3. 突发性[br][br] 网络舆情的形成往往非常迅速,一个新闻热点再加上一个情绪化的观点就可以掀起大片舆论的波浪。[br][br] 4. 随意性和多元性[br][br] 网络舆情不同于传统媒体的一点是网络舆情没有门槛,所有人都可以通过网络媒体发表意见和评论。网民在网上或隐匿身份,或现身说法,谈论国事、交流思想。网络为民众提供了交流的空间,也为收集真实的舆情提供了素材。[br][br] 在旅游行业中,越来越多的旅游者会在网络中表达自己的情绪,同时旅游者的决策也会[br]受到网络舆情的影响。网络舆情中的旅游者情绪对证券经营机构来说具有极高的研究价值。
[b](二)获取旅游情绪的分析方法[br][/b][br] 应用网络舆情分析旅游者情绪,需要从大量文本信息或非结构化数据中挖掘有价值的资料。通过网络舆情分析旅游者情绪的过程如[color=#0000ff]图 7-5-1 所示。[br][/color][br][br][br][br][br][br][br][br] 图 7-5-1 通过网络舆情分析旅游者情绪
首先,应用文本挖掘技术,从杂乱无序的网络媒体信息中获取有价值的信息,把非结构化的文本信息转化为结构化文本信息,从文本信息中提取情绪测评指标,结合属性词典和情感词典,应用情感分析引擎,获得情绪分析结果。其次,可支撑两方面的应用:一是基于情绪分析结果,以及情绪与旅游市场之间走势的关联,对市场行情进行预测;二是基于文本信息中的属性和情感倾向,指导各类旅游营销产品。[br][br] 对于网络舆情中旅游者情绪的分析,主要应用网页抓取技术、特征挖掘技术以及情感极性分类技术等。[br][br] 1. 网页抓取技术[br][br] 网络爬虫是目前使用最多的文本采集技术。网络爬虫又称为“网络蜘蛛”,是一个自动抓取网页的计算机程序。通用网络爬虫的原理如下:从一个或若干初始网页的 URL 开始,获得初始网页上的 URL 列表,在抓取过程中,不断地从当前页面上抽取新的 URL 放入队列,直到 URL 的队列为空或满足某个爬行终止条件。主体爬虫的工作流程较通用网络爬虫复杂,需要根据一定的网页分析算法过滤与主题无关的链接,保留有用的链接并将其放入等待抓取的URL 队列中。然后,根据一定的搜索策略从队列中选择下一步抓取的网页 URL,并重复上述过程,直到满足系统设置的任一停止条件。有别于传统网络爬虫的是,主体爬虫主要解决三个问题:一是对抓取目标的描述或定义;二是对网页或数据结构的分析与过滤;三是确定对URL 的搜索策略。这一过程所得到的分析结果还将对以后的抓取过程提供反馈和指导。[br][br] 2. 特征挖掘技术[br][br] 特征挖掘技术是一种能够从结构化的文本信息中提取出关键属性词的技术。属性词一般由名词和名词短语组成。产品具有多种属性,也称为产品特征。一般情况下,一篇产品评论信息可能涉及产品的多个特征。产品特征可以分为显性特征和隐性特征两类。显性特征是指出现在语句中可以直接作为产品特征的词汇或短语,而隐性特征是指句子中没有明显的特征描述,需要对句子进行语义理解才能得到的特征。提取隐性特征需要自然语言的完全理解技术,而该技术目前还不够成熟。因此,目前的产品特征挖掘只考虑显性特征,在网络舆情中也只能识别显性属性,进而判断旅游者对不同显性属性的情感倾向。[br][br] 3. 情感极性分类技术[br][br] 情感极性分类主要是分析主观性文本、句子或者短语的褒义或贬义,即判定它们的极性类别。情感极性分类是有指导的机器自动分类,一般分为训练和分类两个阶段,具体可以分为以下几个步骤。[br][br] (1)确定情感分析单元。情感分析单元即情感极性的分类对象,它是由研究目的决定的。情感分析单元的选择,直接对文本信息的情感分析效果产生较大的影响。[br][br] (2)文本表示训练文本。文本表示将决定选用什么样的文本特征来表达文本信息。就目前的文本分类系统来看,绝大多数都是以词语或者词语组合作为特征项表达文本信息。[br][br] (3)挑选分类方法并训练分类模型。已有的文本分类方法有统计方法、机器学习方法等。在对待分类样本进行分类前,需要确定分类方法,利用训练文本进行学习训练并获得分类模型。[br][br] (4)运用分类模型对测试集进行极性分类,评价所建立的分类模型的分类效果。[br][br] 情感极性分类算法可以分为两类,即基于语义的情感分类方法和基于机器学习的情感分类方法。[br][br] ① 基于语义的情感分类,是指通过文本信息语义分析的方式建立情感分类器,主要有两种方式:第一种是先从情感单元中抽取带有情感倾向的形容词或者动词,以及和这些词具有修辞关系的程度副词或否定副词,将其称为情感词;然后对这些情感词进行情感倾向计算,并得到它们的情感倾向值;最后对情感词的情感倾向值求和,得到情感分析单元的情感倾向值。第二种是建立一个包含情感字典的情感倾向语义模式库;然后把情感倾向分析单元按照这个模式进行模式匹配,计算出情感倾向值;最后对这些短语模式的情感倾向值求和,得到该情感分析单元的情感倾向值。[br][br] ② 基于机器学习的情感分类,主要算法包括朴素贝叶斯算法、决策树、人工神经网络、K 近邻算法等。对常用文本分类算法分析比较发现,支持向量机、K 近邻算法、朴素贝叶斯是三种较好的文本分类算法,其中,支持向量机具有最高的分类精度,但分类速度最慢;朴素贝叶斯算法具有最高的分类速度,但是精度最低。基于语义的情感分类算法和基于机器学习的情感分类算法各有利弊。[br][br] 基于语义的极性分类算法能够更加接近现实的语义特征,但分析效果依赖于对语义模式的正确归纳;基于机器学习的情感分类算法,直接明确提取文本信息情感特征项,但分析效果依赖语料库或训练文本信息的代表程度。[br][br] (5)使用获得的分类模型对待分类文本进行分类,并对分类效果进行评价。[br][br] 文本分类中普遍使用的性能评估指标包括查准率(precision)和查全率(recall)。查准率反映了一个分类器对类别的区分能力,查准率越高,表明分类器识别的正确分类数与总分类数差距不大,即识别的错误率较低。查全率反映了一个分类器的泛化能力,查全率越高,说明这个分类器越能够把正确的类别识别出来,但并不关心识别出的总个数。[br][br] 为了判断属性词所在文本信息的情感极性是否符合人工标注的真实极性,可以归结为一个二值分类,评估选择使用二维列联表。判断情感极性的过程可以通过列联表进行展示,如表 7-5-1 所示。真正属于该类的极性数即在人工标注中得到的情感极数。衡量查准率与查全率的计算方法如下:[br][br] [math]\text{precision}=\frac{A}{A+B}[/math][br][br] [math]\text{recall}=\frac{A}{A+C}[/math][br][br] 表 7-5-1 评估极性分类性能的列联表[br][img]data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAwkAAABsCAYAAAAhSv74AAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAAJcEhZcwAAEnQAABJ0Ad5mH3gAAFc+SURBVHhe7d0HXBTH28DxHxxdUASpCiIqVuy9G7vGHo1Bjb33gh27WFBjrMQeezf2XmLvHQWRqqIUAQFpx8J7B4cBBIEk/zeo883nwu3e7bl3u7P7PLMzs2phYWFJampqyGQylH9TH0rq6uofn6f+VUr7nrQym5dRTt6Tmb+7nCAIgiAIgiAIuaOu+isIgiAIgiAIgpBMJAmCIAiCIAiCIKQjkgRBEARBEARBENIRSYIgCIIgCIIgCOmIJEEQBEEQBEEQhHSUQwYlpTwVBEEQBEEQBOFbl5SUJK4kCIIgCIIgCIKQ3scrCffv3//k/gefu09CZrKan1ZO3pOZv7ucIAhfjlKlSuHh4aGaEgTh/5sog4LwbVMeA5TSXUnQ19dXPRMEQfj/ly9fPtUzQRD+S6JSThAEJdHcSBAEQRAEQRCEdJTVBcnNjTw9Pfnw4UO6JkaiuZEgCP9rykuayofySkKRIkU+NnVQHn8EQfj/kZiYmPxX2dTg+fPnKU0NRBkUhG+K8jiQtrmRMvJOThKUJ+aYmJhPkoHUg0TGID2roD0nwXxO3pPR31lGEIQvR8YkQRCE/39pkwRBEL49mfZJEARBEARBEARBUBJJgiAIgiAIgiAI6YgkQRAE4X9MevWYx28k1VSqKK7u2Mrtd6rJ3Ip5waMnoaqJXJAC8fEJJePa5Ib0+inPAjP5hHh3btx6Q7xqMudieHj5Om//yUr9KyQ+fIhWPVeRv+TObS/FGgrfGumtNz7hqokcyjNl/V8o50qirH/bRJIgKMpKNMHeD7l25iBb1yxi6rARLDz98h8fXLIjSZ/7F/7zI4gg/HvUbjC7szM30p19ZCQ+Xs+6y38j0FeSJXB+ck9WPPHjwu9buPQqh6drmYzbS3uz+NoH1Yy/Ieky07q4cCdONZ1KCuFQz0bMuZLL02zcU4459mbZ1X+wTlmKxufiSW7mKCqJ4/5vk1hxMxIpPIzw5EUiub5oDL/dz304JHyBwk+yYtmfBCueSu/2MOTHVTzJzabPK2X93yjnSl9MWc9NOVcSZT0nRJLwTYrkyb4FTHCcxsy5zjiPaE9H53N4hWtgVqY+HQcOp0ctS8Vh7X8pkuMrVnM33fFFkdkHvuDBpf3M+b4sHX97glz1iiB8yWQFTDCxK0xRXdWMZOroF6xH9SoFkqeiXz/HOyQXybFWUapXtkSubkHd2nGsmbmLt3ExxGT7EXqY5NcjQZb28C8n8o0Hd/88xs5FvWne93c8PlP4ZIYmGBUuiLmGagbxxCnPqzJjClnaUdhGK2V2DsU9O8tl05H0rPn37pUR9/Ie504e5Y+929nkupwlLi64KB6LFi5gvtMw+g6ax5YLL7Ku9ZS/5MbFR4ShibmBFlGSBjFXlzJrjz9SUiJJapZYWPxvj4hCHqFbiKD1izj8RlHELM0xzJ8Ps1zsznmnrP/zcq6Ul8r6Py7nSqKs54pIEr4B8shgXr54zPWzB/l9xVwmDBnK0mcVGL5gLjOnTWFIwyJoFqlF+y5tad6oNtUqlqJw/v91IdHANOIo44aMZeLUaUyfMZMZM+awdPNxbvvJkelYU6VycUUxFoSvgTpGRSwwUDx7d/8P1q1Ywvx5i9h68TLHVsxixvSpzFmzm9MPXmZxgovgxK4DpLZiiAt9ifud27yJ9uHQr7OYtuouhB9iTDcHxmy4p0jBP09dEXQULZqI5+GFjB87hak9G+Hg+oRQDXMs1NQp064FJT5X+JLiSMpXkAKqw4Tks5HBA7fiqTjhappZU9gwN8ePUC7uuEvNib0op62alUwiShHQ3Dm5jAFdZ3NWWbWbBW1TE/SlBHTMSlOrVXeGjHXE0dGRCWN+pIxcnTrz1+PyUymyDGc0LbEK3cpwl1uo6ZhRyDiBRy8L0qRhYWTyt4QWrU9dExE4fBM0DDEqXwzL/MrnmuQ3NiRdvJ+tvFPW/3E5V8pDZf0fl3MlUdZzRTmuqBgC9SsX53uWTbsfIbPW4/65aIauGkt5bYngwHcYmhlwb4kzz3+YSU/FwcT/9A5umranSyVD1dL/KxL+68axvpQzsxvoqealiuaU4ySCJiynp4lqlvDV+zqHQI3H94kn+Yp6sqD3KYybdaZb1zJIIeqYFzfixZJpXKjXnToxvgTIKtCqkW0WAUkE23tUZ59ZB+w0QKZrhIWV4gSdeIlDQb1YMrY6Om/9eW9djMLZnt+iuTR3Me9GT+d7zTjUFSfrZ8tG80eNpUyr+ZaNU3dTevok6mQslioRwcFoaJ9iUCd/Ru2ZQjUjxSeeGUO7U23Y51yUrUM3UH75AhpnsXxG0Xdd6DjkITXbFiPhg4SahgYaMsXxXnHM18hnjIVpEIcPmeKyewSlclRrIBHk+wqdwro8XPsr14r1Z3TrYqSLSTIT85h9R6OpFn+Vc+XKw41oipQpSdVCV1l/rwWOPYv+j6+u5g3f/BCocd4sH7gMu9XLaZm4nb59XtP+R2O83dzxehWIVKYP00c35tPK5rxW1v9ZOVfK22X9b5ZzJVHWP0sMgfqN0bZpyuCJYxnQqTpF82uimbzVE/DdMpG+46ez8tYLRSGbwdSp01n8ywIOPY1RFL//NRn59dUVialqMlkcIe63ePAyHnUtffR0VLMF4YuVgN8Lb+IU8ZZu2bZ8r/M7iw9HE3pnN2tXrOXQXW98PbwJTTLC0khD8W6J+MwKn/SBWMOODJ+zkNkTB/NDbUtivf1Qr9UC89CXqGmFc3rBOHY+y/pCe1yID09uXuDIzi0cv3WRbYrPGTR8Ecd84gkIyYdtcV3kXmfxMm9BxXA/Xmdsg5wshieXr/I6QR2N/O84OGoAK66+4endWJp1r4+yakFNSxvN5DqdeELevCVK+TQr0Q/Ysl+Tqbtm0qP3GJwWODNv7mxmzZrFrJkzcXIcQZ8mxShSohRWOb6sKEP39VmWzd2Au/1wxmcXOES6cfCXmcxcfAwftw0MnL2PC0d90SggEfvhPZ63wzEo+o6bZ/9g66q5jB02kwOeoiHkVyPeiwNTutGpW28GDx+L49zNXHudWggTSQh/j7xQJdoNm83K9VtYMy6zBEEpb5T1f6ecK+X1sp7Lcq4kynquiSThm6WNjYUpNm2ms/m3FUyePpd582YzpEVDSlgb/r9k0TKtGO5smcGEiZOYOs2JaVNns2zvFW4/8SdWcXAW146EL168H8/85eRXU97NVpNiP/7K7Jb6mFZpSG37ohjrapAUHYjXk9tc/GM1k7q14ccZx3iVMXiQggl978nJ+XP4Zcs5nsdZUKdze6oXrU1l6ToXHu7lWGJnOpfJ+kJ7QpgPz17KMVUECcEeJem+wJWN65xoZx2Cv2RKSYO3nHb9nXveBxnRqTlzjgVkUlnwgVevQpLLpqxQHRydmiD3vM7pF/pYvTrEhpUbOH7uKKsnjWPcmPHMWbqLq/6ZRUJKIVxadwT9n1qhF5aP4lZGKTWr8hecOXCJV6pzc8LbQCTjQrk6JhlUq4ypvh1t6lqkWy4uxIM7D/3Tj15iUI6OYxSBw5Qe2CXo0ujH5hQzCicg3ICy9SzxDYhE45Uv7zxOciiwGiPnT6FTyRxnLEJep1WcTs67OLBrM64rl+IyZwTNbFP3GnV0SlSnZeOqFDfJvN7/ozxS1v+dcq6U98t6rsq5kijruabc/qK50bci7i4LJ1+jw8KUS3mRe2eyzGIiE3R/pdu6EqxZ1Z6wlY4cqruQSdVydNHuH/lwZDyTIx1Z4mCWoe/BB45NmkP81AV0VDbsFL4JX2Nzo9Azjgy//RObRvgwe1F+Bnz3nA3nQtE3s6F06WJYWRWlqLUlhqF3uBpWigb2WTTziz7O5JkRtKmXQJiaBsFeDzlyTpvxu2dS6tQwWsxVY9y+X3Eolt3pVY7P5lH0dvWkUKUK1G03iEH1nzO541U67ptL5cjXJOjcYeZwPwZtGYN9xsNA4H6GzYtn3twkJk3QYbFrJ5IuuLDgcTEqFjGkWDkjrq4+Ru1FTtT87CEkjucHV3BK/yeGNghh+aiVfChuh3ZiIlKCJ3/8UYj5pxbQ2EhxNDg8ktGhU1jX21y1bAZSHJFh73gXFkZ4eDhhYeG8Dw/m7rETRBS3x0R5CkuSiI+OIjI2EU2DkrQaPIDGlml/qygebV3KadMeVLu1jtcDplHl4mqu6+XD40l5xk1sQOLukcxSc2TlT1b/L5Uo/5VvtrlRXBgvXzzn+Qs39q65S4f9q2iZuJXeXdyo3UKXFy/8eBlsSLtZ83Eo8+nOnbfK+j8s50p5raz/K+VcSZT17IjmRgJSmDf+H9SRJYK6tiYaWnoYKEqDJGmio/P/s1uo6WgokgOtTDonJxIbH0+6QRkE4QukoVmBtu3KoZ1cF6OOaeNhzJnVB7NXF7h3+xInD2zlt2ULmTNzOAM23CQiZbFPxPt7oW5VHk2/F7w0Kk+Tbo2oVrM6FfUkohIMKFRIDS1VeYm5upNdT7JoQxB5g/0vqtCzSS36TO+C5vWLPHv0FI3ySVw5E4CBlTVJV66j1rYTZTM58YeeP4S3TUX01BTlV6Y8YPhw0q0wPw/+ga6dmlK9mAEfB0HJksTrK/u5W8iBoc0KI9PIj36RqnQd7YjjxIk49qpN5UaNqakIGpTvDQqIpoB55rUF4Y934zzJCZe1Ozly8Q7uL98j6RTCpnxtGpazo7LDRKZNm8Y0pxnMnu/CL78sYdHswRkCh3Ae7lvPVdOejKz/htd6dahYQI0ibQZRIVqbYtJ5zr5SvOdFAZo1+TaDhq9d3PM/WDRrKdsvPiNM0xxzwzSVktqFKNd6EDNXbWTXnqWZJghKeaqs/8NyrpSXyvq/U86VRFnPLRGGfUtioojwusRO1/XsvBKMlkESiVISCe9CkZmZKAp8AlEf5Gj9P/UFUNPVJsLjIscP7mDjioXMmDSawT/3ZfqBF0QqjnvZH4AEIW/L36gnP5VXnoUVgYOa6nArK4COhhVNxkxkxM/f07RdP8b3aEr9YjZkXvQkgm+/x6R+URIiwwjyusXhzQd4EJZI2KM97I3syI6V9bk+zYkDHlEkJfjwwiuz9spxuB28julPnTCMl6NeoBYj5vRC4148VcYMo6znXq68e8nZOwXo+H3mHfcKNJrKrz+njB4iU2TxSWFR2LXsTKmPLR8SkT685v7RrbgumsbMTbf5dGT4BLTsWtG1vmqYZVk+RSASQ7SylYZCvH8AmjbFVJUHcfi+hCLWmTetMLT/kSkui5g9ZRwjBvSiW6fWNG1Qk0pli1HYWIO42Oxqw+PxvXaZd2V7M6RFMfC+z7XIfBQKvsgvM7aR0KgL/fpW5O3BLVyRV6ausWox4auibdeBCc5zmDSsNz+0ro5V/tTQSJN8ViWoVMqc7AbszDtl/Z+Xc6W8VNb/eTlXEmX97xBJwtdOesvVzXMY+VM72o0+jc2oBYwb0Z8ebaug9taX93JFMQ6JxLiELdrSezw9FMlDvv+f3UJdR5t3L7yIyVeUKq16MX7eMly3bGR2+4LEJWioOkQJwtdAeVZU+6ufzfvb7PtlESt3nufu85dEKc5x6jJZ5ids6SV/hhShdVkttAtaUsy+MV2HzWNev2K8jSxNn141MS7Zlfnzm+L3y8/0nL+Dk/eeKk656clfHOeSThu6lI4nOl4tpXzFuXHtgx2Ni1jRurUOR+Yt4VHRTtTPoiWEzKIUdoUUa5m6voXsqVgsjGtbl7Jw0ULmL1zNcc9Y4pNMKPtdd3q1Ko2+atm/aGNiWjDNd9VC9t6NMzt/Y9mCmTgt28+tl/4k3+hWcUzyC9KjmPWn1xs/Tw0NZWCjCkaypoVVcVNeH3bBac4Cpi+4jKaRD8c2bsPdpi4VzfOjWbgFtaP/JKZmY8yyiqiEL198GC+fXOPkXkUQ7a1qW6+m2DffPebo9nX8MseR/p0bU6FiZ1wufWY83v+4rP8b5VzpyyjrOS3nSqKs/x0iSfjaycypUL8hbUatYvfmeQz4rnjy+M3KGpICNiUwVY/i5n2wt/TifrA2NtVrYZM6IPL/mrYWJqXr0bp5XSqVME9u7pQsMYpoSRttkSQIXw1lTVeaHVrbDNuSRTHRjCTgTSQJikOxTHlJPxORdy4SW7Ut+Xz9KVK9JrYW1pgZFETbbT0uxwOSO+dJry+z43gULV12smruPPpW1iZdF0K5Dxfu6tOqYzl0leVLrpF8pS782mXk1ZolD6WoW7wqBR89pkCl1Jq9z4iOI0Fdtb6KY0ydnmOZOGEikyf0pFGlyjRv15IG1cpQzNwguSbys+L8eBWgiWmpWrT6cTDTfr/LjQVNSR79OM4dTzU7smjh8RlJxMQmIMvB5UiZWU16TprHnCGVMKkygBmDG6AZX4EB3SumdK6MfsbdQA1F/Ob7aUdI4YsX73WIOYP7MnjSr+y7/RaNojWoYJVaFhUhUtwHEg1LULfzcBbtuMCjh/txbPC5sbn/w7L+b5dzpTxd1nNezpVEWc89kSR8AwyKN6BFLatPxmQ2aTOVwYY72KfWhs5VC+K3cQ5njGpTMdcn5L9HXU2dBHnaS6VxhHopDhpuPgR/0EVP7J3C10JKJDG1CYLixJakng/TktVp0X0MTsMaY6yIKdRTT8RpRT/mwusydKgZxLkr3uSziePO1lO8kWQUbT+U5okevE2QCJHZ0amaH7M6j+NCvub07VCJdEOXaxaj+Y/NsFFGBYmxxCZqoZXkw6mHRjRrWFCxfgGcWnUQY6fRyHf8zsPolMWyIkVHE5/Z+v4dGmaUqVkd+yoVKVXMHEPd+I9DI8tfPiHErELOhkSM8ebmVQ9VW2+J8PB4dJQNqnNIHikjf+INFgwZxj55cQrKlSsRxf3tR9Dq+Sutw/ZzyDer0VuEL5VW8fY4uW7EdelMxvTpRNOq5mgr4/xkCWjY1KFdm8bUKFsUo5ycG//Lsv4vl3OlPFfW/2E5VxJlPedEGPYN++C2l2XHTBg+qQUmeiXpMHECjbwmUq/dTI54ZdHx8V8UH/iaa4dWMnHsOMaPH8fY0RNx3niCa5cvcPuDAYZi7xS+ElJoOJGptYtSEKEfjLAqVRIrE93kS/FJCfEkZnYi1rOnXafKvD+4igfaFTEo1JRuFe4za/ouHkfZ0X/+WKprx+Nz+TL+5YayeLgGVy+/TX8VIaPE90TEahB36QQR1TtQVj2Qq5u3EtBoPAMatqZnjbs4jtnArbdZjw+eEBFJrIZG8jeKj37HGx93Ht66zNlTN3j26AKbFs7EadI4hvXtQmuHhfwZkrJcpmTmtBnRlyqpAZgUzo1tqzjiGcXLPx+iX6vSJxUcmdK1xd7sJfvX7ua63yt8XyVhaJijutJkmtYVqGCpR9HOS3BuGsLOVfu4fmoT542707uaOfWa5+eIy16ef1AtIHydYj3wcE8pQdK7QAJj1NNeF8hWninr/0I5V8pzZf0flnMlUdZzTrndk3NmMQTqtySMx8cOciXcjg7d6qW/MYyypmFaF9ZZb2f3EJvM203+S6JvzKHnwVpsXdgsfa0nody+9orydSrkLDgQvgpf5x2XU4Tv70uHR4M4Nasm2uH76PP9QwaemUNtfLiwfR83753jRaM9rO+aX7VEGtJLNo5ejvlMF1ond6aTCL75O4tdLxFb6jvatmxIiYdr2VpiKk51s7/1adzDxXw/6gUtBvahX/tCPDh0nfiqHWlRStU1M+4pm8dN4mCYDdXr16NOgyY0LGuc7lgQdWIkHa52ZVuHl+y4EIReIRMKGRtjXMgYExMzChUsgH4+bbS0FI+/cRCRgk4z1WEyJ+IasvDIUlp+pu10RnL/UziPn8ExgyHscu2Fbbbxg0TIkzOcuhmMSb2ONC+lbFkdg++F3ZyOqkn3tmVSOq3KPdk6bjR7QktSu2EbfnBohl12vVm/QN/8HZfDjzO45Um6XFhOzdN9qXrmJ+6tbJZtx+VUeaWs/xvlXCmvlvXcl3MlUdZzIu0QqMrI+2OSEB0d/TEpUFIG5mkfaX0uaP/ca0rZvZ6Vv7uckIb8NXfPP0Feqj61bLI4yMhf4vvGFBvr/3W7o2jeBHzA1NLkf5qMCF+GrzlJIPgCfzwsTdumFsikt5zaeQO7nzqQPNR5vC9/THLGr89KRtln3qpXkqRM2jHH8PrOSQ4dPc/1hy+QGsxg/ZhaGRLuTMh9OLjzGbW7NyDyjhsa5WtSLJcnQMl/NysvVmKIavST/4WwS0tY/KIVM/uWzVnb6TSinx3lWEgNOtU3zf7YEuPNfXctSlUu8vG3i/G+xo1oOxqUz91N3L4G33ySoAgkg557QXE7TCMus/eSKe3b52I/zytl/V8o50p5uaznqpwribKeIyJJEAQhz/mqkwRB+EKIJEEQvm1pkwTR6lsQBEEQBEEQhHREkiAIgiAIgiAIQjoiSRAEQRAEQRAEIR2RJAiCIAiCIAiCkI5IEgRBEARBEARBSEckCYIgCIIgCIIgpCOSBEEQBEEQBEEQ0lHeeEAMhiwIgiAIgiAIQrJ0N1N7//598o3UlI/Um5Yp/6Z9pJX2pmsZZXxvRtm9npW/u5wgCF8GZRkXN3EShP+OKIOC8G1LjbXFzdQEQRAEQRAEQfiESBIEQRAEQRAEQUhHJAmCIAiCIAiCIKQjkgRBEARBEARBENIRSYIgCIIgCIIgCOmIJEEQBEEQBEEQhHS+vCQh+hGXrr5BUk2mkPPS248Y1VTOyHkXHKZ6/m+Q43X/EUHpV+xvkON+4SzuUarJf0QiwNc/V7+L5HOb2/4Zv0QEp9cs51KQajK3op9y626waiIXpAA8PIIzbOvckXzv8yAgk0+Ie8j5P/2JU03mXDQ3T53j1T/ezv+URGRkhp1E7s2VS88UaygIgiAIgvDPfJIkZDY+ck7n5dQ/GoNZLYLjY/qw6plcNUNJHbW7KxnqfI43OQ7eYrk8oxOjjiojX4mggEBFeJ6BFMyzJzkPsmNPDKPWiMO8U03/PeroPFtO+/GHyF0KI+F/dDvH0kWv8bxYO4Jx+/1zHmirXWB49dGcTxdpapB0y4WFp/5GoK8kk3O4dyNm3n3B0V+Xc8Inh6G5TMalKU2ZdDZSNeNvSDpJ/xoTuZJxI0pv2dqoKCNO5TKkjrnHru5NcTr9D9YpU1F4HNvLhRxnHzFcm9+HmRfCkULfEZq82HvOOnZj/rXcpz6CIAiCIAhpfXlXErSsKGJtirG5ZvJk1LtQRbgko0i77pR9coJHqvgo0u0KN19+LuDSwdo4gaBoHcXzePy3u7Do4GF2bVjOAqdxDBs8kAH92lLefiqncxT1a2Jta8mH9/FoqeYQ5Yt3ZrXYnyXDtERJ7AoZK9YwNxKJcnNl9sZHaWrHk4iJCiI2SVvxqTkjMzLHooINJfRUM5KpYVCoOQ3rGCVPRfk+wT0wF99LuyQN6hYlXt2KZk1imDf0N17GRBOd7Ufkw9wwH/EaaddeTrj/Y64c38Uax2bYtfiVR59kd3+RGZtjamNM4Y8bJY4Y5Q8kM8Pc2h4bO+2U2TkU8/AQJwvPYkQjA9WcnIvxvsqhvTvZsn4VS51nMHnCBCZOnICj43jGDuxE8+9Hs/zI089f3ZB7c/7YLUUiqo1VAS0iJC0+nJ7C0LVeSImJJKlZY2WtoXqzIAiCIAjC3/PlJQkyXfQKFaRgcmwXw4NzZ3nq/YRrf3pTqHYhbjlPZqLjCH5sW5/hG258tumFlm5xbIookw0tDBPdOX/7PTpW1WjdZxIurmtxndKJHmP70cw45f0ZRV3dxoYrgR9r6dUKmlDezPivgDzhElOajeFkuGo6h9T0C2CSGMSNE3vZvMKZiYMd6Dp8HXc/+zkxeL4uxs8OFRTho4oUwLOQirRvbqaakRPqmNhYYah4FnRtK4tmTmbsqAmsOHaKnTOGMXhwf0bMW8uB695ZBLNh7HHdTGqLpZhgbx5e+ZOXkc/Z6jSMAbOvwrut/FS3Ht1crpLdTyPTKUnJ4hJu2xzp7tCXfo2tqO98m2CNIlirq1OpRyfKpeSLmUuMJcmgEAVVG0XyWEL7NstxkzTRKqLY/sY5TZ+Ugjm++jKNXUZTRVc1K5lEhCJxubzXida1hnPorWp2BroWFuSXEtAtXJHvug5j6vxFLFy4CJe5A6gkV6fpppNsHWL/1/bLjGZRbIOW03HCBdT0CmNmFs8tb2Pat7ZBFv+S4BItaWaRm+8kCP8OKfAZz3JdKSIIwn9DTvgbX3x8fD4+fH398A8IIuKzNVXCt+QL7LishpqmZkogLgXjfussF84+4G2SKRVb9GL0jPnMHVCVyv1OcsipLukqxDNQV7yqly+l1lVd04hyjTvToXkdKtiaoBfzDp/bbkQYmZJVDKprEagIfLd/DIjV9fQoqHh8DNFkmqjlN8AoXUCZQfh9ts0cyxjHKTjNnIvz/AUsO/SIoPAw4g1sKVe2Ab1/3cGelQOoqozcPyHx8ulTgiJu8Sh/W7qUkClO1v68VGZHEffwLvQd9fOnvBNCufr7Rv7MtNVQHJ537vE2MYkP97czw/kPQkvWp80PQ5i+dBEj2jSmVe9edG3VkAYdhjK6Q8ksg9ngnX3o0LMXvXr1YtiMNRy84Yt6hXLYFXfAZfNBjm/bxrbb9zg8qW5yMvI5ajIDDAoYYNd5Nlt2rGJIvZo0/qEbHZuY8izQnoEdrbK8ShL29i1RSRIxz5/yWPWdY54+wtusKMo4Wi2/IQa5KAFRVxYz5Zg2iUfG07e3ImEZMFiRNCkeQ4YzbeV+bgQE8lpeCNsskkp0bWncrSddWtWjUgljIr1fECZ/wylFovC41a+s7G2PvuqtWZNh8+MoepXSRENdA1nUeR5QGi23hwR6uqNXswXWIkcQ/t/F8XjLaEZsuv/5K2GCIOQRCUSH+HB5SVfK2pam/cR17DpwgD3rFzKkZWWqtx3Oigsvc95UWfgqqSkeyR0E3r9/j7q6Ompqasl/lZTPlQ+l1Hmp0r6WUVbz08rJezJSLhMT9ICVI6byqrQd+TSTeJPwAyu6PWb+6XpM7xHHgd9vo1m2GpUaVadEJlGsPPgFD5964vPCjSubV3O6YC0qF65FF8vjOD0oThMrHdTfPeKBvBy1y1pgXroJPbpWI6WhTQaRBxg8/gOLfuuJMg6PuzaVQRe6sG5qpZTEInQ9HfvC+j/6k1XcqCTFxZGoLfHo8kNs69dG795CJl1uy6JRNlwc34ETzfaztEVWzVskfFwn8Yvfa/yCy1LDOoGgq9tw//4Ue+1W4LC1oGJeOMHhMcTGBHLz0Ft+On+RyZUy/jhRHN9ymDJtE3CeqM2AMq44q61klPZOTgdo8P7sUYK6DKdLcXNMLEpSpYYNupIM7YwBqeTP4i7TKbllM81jPLh36TSHT/hSblBFzv0ix2VrW463b4X7rGssqJp5mhET+By3Zx48f/aY065rcbOtgFFcUXoum4OeyzACZ22nf+giBu+ux7L+prwzLYHNJ4lYNGc37cfi+0jGtLyCdf5wikx2penpERzqtAuXqt7M+nEFtXa40kIvjkD/YHStiyRvx0xFXcd55DGqTHCgiJYJtspEUvVSKslnMZ1n2LJtS6ccBPuK/PDMYqbtD8C6w2jGtbTOMtn5KPwev7us40aEHgYJ7vzhegeLEWPoVMYUIyNrLANOcbVMB+rEevH00R2u35fTfO5y+nz2Ukveoizj/6jPkvDfiDiDY73vWak+huNXFtA4JwVAyJNEGfy2RJ8ZR7XW66i0yY8dPQqmzIx6zJqfWzPyUlkWnP6DcekvnQtfudT4XHkc+EKuJIRzd9NUho4Yx7yNF/GNLUjZjhMYUdaEMo1MOPTbXQpUMEbDqBpdHcrguXYy87Y/zrxGK/YVT9wC0S7VjGbV6tLD+Xd2rBlGWZ0S/Dh9Gb8uW6IIzqvzIaYio6ePp3vjYhRQLfoJWUGMCqb5CWUaaKj/FerJXweiblXk881HFGTa2oqkQkbi1Q38eilCNRcSQy9yXa0nQ7/7XPt3GYVMw/FT78WsKUMYP30qQ5rXo0JZIx675ef7IT/w8xhnVqx2Zd3qsbRpUJ/apTJZo7gXPPCKw1AtUTGhhd2gPfzW2YDCdVrTpFoJTPU1SYp6zbM7fyqSiTn0qatIogbvwidjNUPCW4LC3Ng3biROyw/zOMaaJn160qBEU+pI5zhyYz27EvvS55Mk5S8JIe489JJjUTYfbx+VZ9imw5w6voIexYPwSrCkXP5X7J/3C9fct/BD1ZKM3O2XSW1HJD6+b5OzYJl5UxataE/80/PsdzOgmO82Fs9yYfehnczt7YBDt+6MmPIbp72yqjMJ5OSiHeQf0pV8IfqUSU0Q5E85uOkkPqo+EfGvAkgwM8s+2FcxrFcbywLl+bFZ+gQhJvAxV256fdpUzrAKveatYc2y4dgn5KPNoE6UMn2HX6ghVZoX5bl/BJrezwl8tI9tAfWZvXHZF5UgCF8qiVdH9+BmVgODZwfYdz5UNV8QhLxOTVsLrYyRoL493bo1xejdJf44mU0/OeGr9oUkCYZU7TOP1SuWMHdSd6pZm2NjA+/L9aCJ317ed5zPhEaWyYGWrFAjJq+bQ+Uobz6kLJyOplUjeg/tTbt6dhTUVoSQSZG4X7mJ74dE5CcnM/1CNOqGhhS1MFUEgtGcmjiYtZ9EwqkUAdiH94RFheL37B63nvjx4voW5k0ew9CBAxgwfjNRNiXJWQ6uTZkKRrx9HoJcnkB0kDsnV+/HW/Ml22ZMwnFkf34etoRzmYx+o2FoRBGrctjbGqckJGo66Moe4lGgGtXym2BlolqDpCii85lhkUm/1uDjizkRYYcOicmXloLPrcFptgvrDt7hTaIJzRZvZ83YSTj1bU7TXks5eNWde+u6USxjRCz3JdiqPz3q21G5dH6iPP5g0uC1eGpa0qF1DGNb36HdssGU+kwkbVCuHf36tcXy3gXcjD1Y9bMDU/Y8ITLyLrcuh/A+3oLm865wcUUt8pWcgtOPRT8NzAOOcdxNH3PFd9WQqaNRfhATSjwlskZjNONNqfHTADp36cXs33ewY9c+9mybww/FM1upGJ78/gs3Ko9jSPkYRbIwmDETJjDB0ZHx48cxYNwunqg6ViT4+ZNoWTTz7S3FEB7oh6fbPW5cOs3RA7vYsu02fnfW4ThqJCNHKh4jhjG4Tw/6jJrP2k27OeeX2X4Xwa1VS3BrMZummrFU7DWdn41vcPLYLm7Qgd6D+tDQRAuLilWxTem4Iwj/W3J39h+Bn5aNpH1hPw7tO4HomiAIXzKJmKgoRXIgI1++fF9iu3ThX/LlbXtJTlzIXQ5tuc4rn8t4l3CgpcET/tiygvmTu1OryTSuaNZWBF3tM28ilEoKJ+itB9f3H8THuCTqkeqYF9VHPV7xmoYWxoYFFMUjntj8xShtlnlEKwU8wePifmZMWcDGgxd58CoKXZs6dOo/kYVr1rFy5lQc6lh8GsB+FInbwaXMmDaD2c4LWP3AkGJhy+nQfhoHt67hUslBTBzYj/GzFuCyfD1bVo2jSZFMPk1LkazEx/Ih4BEXr70gLkmDOI9QirZohPzEJMbveZEyvKsUSZy+BWaZVC5raNXAoXsVRYCrTBHUsWg7nbWuYynsc4yrl06yb9NynKeNY8TQzrRafCHL4VnjvJ6iblsN7RdP8TGpRodB31O/cQNq6ktEyAtgbq6Gtmqviz69ht/uZjHAbPh5Nj2ty8j23zF2RX+0zh7lwa17aFaF0wf8KGBbnKRTZ1Hr3idDJ+IUwYe24W6nSCIUeaC6TPGbSR7su1uMUVP6MaB3exrYGZL9GEASvqc2c9l8CE4dbZBpGVLAph4D5y1ikYsLC0c3oW6btjQ2SXnvG/8PFCzy6XWn0NtrGdNrEJPmu7L96GUeeIeRoGtGyWpNaFXFnjpDl7B8+XKWr1iF66Zt7Nq1gy2uU2hbNOO2DuXmBhfOFB7JrJb++Oo3paaxGjbdJlMzSpdS0hH+8FG8x82Ijh1sP7PfCcK/J+raTi5ZdaF9uRb80KEE747v47DvZ4YcEwQhz/mriVkMb+7uYsHaM6jXGMSwziWz7JcpfP2+oCRBzpvrm5k+ai7nYqvTvn8nmrb8iQYaNzl1NwSDMuXI9xY6O4+l3mda58iDHnBw8XC6957PuSBLav3wM63KaBAQK8NCMwn5Bz8ePPLB9+EhfnFyZHu0PZWy6v1cYjD7H59j8/JFzJoyloHNSmFhUYIyxc0xUERo+rU7YflgA5ezHELVgHIdxzJr7iymT5nEhCmDqBETRoVx4+k/fiAFr1ziXX4z9LMroWqJeB37hUW/X8TrTSjv4tUwqt1GkVDkp3b/riTM7MiMSzFIIUFQ0DLT5k8F24xkSDVltK04UKQ2mZIVRFezGB3mLWbGKAc69pjAguHtaVHKLosO4YpA+VIoli2LEx8eQoD7BbYt28SNEImQW2tZ/743lw+05NyAAWx6FEFiggduTzO7kBnDvS1nsRzSG6P4OGRG3zFz7Sg0r8VRZ54Tld3WcyrIm4OXjOjlUCLTYNioza/sHaUc6UkNmSJJSAyJwL5Lb+w/fvlEpAhfru9cgbPjAIYsvcSn/bnj0anQlYEtVFcqZPkVSccHohKSX1QkRP5oliyl+j1jeO6dhG2JTweuNao+kF+2bcF12XxmThzF4N4/0qFVY2pXtsNGkbHFRCubeGUnDs+zJwisMoapne3A4zpnww0we3OMaUNWEt+mH45ja/Ly9185La9Nc1PVYoLwvyQFcmKXJ7UcGpJf8V+Dzj9Q4cM59h18JpooCMIXI4b7O6fhOGUKE4b1p3fvMfzmV5GeQ7tRQ4yW9037MpIE6SUnZzgwdI86HWYtpFtZreS25sr2+GY1f2JI/2ZoX9hHaLcljFWN5Z+5UB6dPcYDnU4s2LgQh4r5Uz4n5jleajaYRofheXovl56HoWFZle/7jWbWz7UzDaqV7dFXDxjENvc0NWaJSYrQMy0NdLx+Z/mxQNX050Tjtnk2+0xG41TPkCT9uowdbMieudt4ns3ZVl1Pn9ItRzBj8kj6da6GgaSGdmp1vVEzxoxvQwHFZPyrN2ha2fx1H4dMKb+B2l87Rugl1k9xZPaaw1x57M17RQ6hDLozrYWXvDn+1pYfK2ujZ1KUUtXaMnDaBjY4luZVeEXGjm6EaTlFgrCxAy+mNqbx2DXsvXpfcXhKT/50Dyd0f6J/xVg+xKmhqdxIMfc4E2XP98Vs6dZVl+2jJnOrZB9aZbG5Zdb2lFdeAVJ2vNFQrK9ZdWqWCuHsiik4OjoydvwcdrtFE5dkTpX2wxjTtUImnZZ1MbdIM6StYk/QCLvLQdf5OI0bwoBpG7no/SLl5nlSKJ4B+tgVz02dizqaMnWScpIjKP5t27KF8ds2gYEjxzNo7Em0TJ+ze/FKHto1p5ZVQTRtOtMk8jhRjdphKY7rwv8Dudd+9gVXpaaWN8+ePcO3QA0aV5Rzde8BsrpIKAhCXqNLFYd5uDg7s2jVdk7ducPRQTrs7tuMjjPOKyIn4Vv1ZSQJMitaztrLwV9+pkomwwQFn/mNE2YjmNzC/GNAJ/e8yEWPjJe8jajqMJVZw7/DSlMDPV3d5CRBCrnHewM9XiT8xBrX6YzsWhVby2KUtKlAmecL6L/Jm0+a2GpaY530gqfv/4rwpMSEDO+ToaZlT9Wqn234pBDFk62z2RTbnXlDKqMjScmfo12uD2Nr3WDqzJOfbeOrrqmF5sdLhRLh4fHo5EvdtDJs+y5gYj1dYl9GKlbbNE3Qm5kkxX8pKVgynSKULl8SC+33+L0MJ0Gxy2iku7nZX8IvHSOmngP6nt7YNGxEaesSWBoWQufuIibs9k/uIyL5nmLN7gi6bL3CgfXrGV9HJ/1vJn/OkSv56dq7CnqJkUTFayUnJKHnTiGv1wkbxT+tV6YehW7fxqh2qewvg36IJeHjlRErmo5wxsXFhaWLRtKmdh069ehCy3qVsCtimHkymFaMJz5+2lhW+I4ug6ay4vx7QjZ3wDz5tUe4qdlTOWcdUFQS+RAjRyOHeYXMshEjlqxn7dSaWNSdxOoprdCKq8GEYTVTruxEPeBKgIYiV/P47P1BBOHfEc3tnWdQL2/As8uXuHRJ8bj6CssaNdF+sJd9aQZhEAThC6JtTbMJU+hTIYbr67dyLpf3ehK+Hl9GkpBB2tHZpFeHWHW5GIN6lEkX5CWG3uGGW2Zdl/+SmKgM8BMIOutFgSalcDtzipcZ8gr1pA8kqhtmElhrUsisOEbKKvpUCYlIadZNGbDHxutjYvKZKFAK4OIqZw7IHJgxuEbKfQMUyUZi8ufIsOk8jX4Ji+jjdBSvrCI/xftf3zuI66KpjBo8nuWK4DVf/oxrHMczjwSs7LKJSBMlReiaumyS4rvrY1muAZ2HzWPF9O8xU+QP6uqZXEeIus1R30r83CiAw6fc0S8Zy6Xl+/GXZJTo6UTnxEe8ipcI1LCnV/0XDKvRnaP6nRn3c+30w4Vq2tFpYEdKKlczIZqYRC20kzzYd7MQHdsoMkTJj/1ztmC2fC7xq5dxMyplsaxIys5Xsux7H+SIVmEqNa5PtTo1qWBXBCO9OKJV20TufYe3hWtgm13AH+3OhdOPVX06JEJD49DVz10xlL/XoGDiOca278iG+DIUilOuRATXVu9Ae+QefgzZxBbPz2SVgvBvCD7DrifVGTNxCIMGDfr4GDahH98XesEf+04TpHqrIAhfGFkBCuRXnJvU1NNWGwrfmC8nSZD8Obv9AI9ClbXsamgk77Vx3N54gnxdOn5yAykp/DV33TO7K7DEq2fuBEty3ocrI0x/Dnla0KGOBSYWeuikjfkjXvHohTq2pTOvHtbVSST6Q5q2IsoAW4oh0NuNu1fPcfzgHg57SJhm0adBHnCFLYu386rGeKY7VCC1K4UUGUmk6jkyc1pOX0gb/8UM7jOSuSsVWb1H+shYHhBASMGyNOvhyGLXxcye1JEyqRlT6EMObNjBtZde3PTWpozV56NYKegd71XPkd4QHGVK8YrlsDXPl1yjnyiPJ1HZETgj/er06FObsC2zuaFTE0OzDgyqcZXhg1y5HVEex03ONNCNw/3kCbyrOLFtuiZnTmZzo5akUMKiNYk9sZfw+j9TWT2AM7+sxK/NAia2/pGRja7Qs9ti/nyVdSfJ+LAIojU0k3f0uKgg/J8/5Oafp/hj/zke3DzC0vFDGdjHgY4talK2viPHP9cyTFaEbjPHUzd1d5BCOL9yNjvcIvA6fpP839XKoq9GGnqlqV7Ei00L1nLuhQ/PvcHYOLvMIj3N4jWpbq1PyT7b2dgxENc5Gzi3fylHTIcxpp4VLTobsmPCOp583IkE4d8mx33PFl7X7/zJ1TNZkdZ0aG3O64Nb2Je2OaYgCHlOUmwcsZ80eZXje3IDu2/JqPpzN77L7q6nwldLGWon11l/CTdTS/A5xAzHFZyMacTag05U05YIPLsc54PeSIqPU1dPJO5DtCJwDyf4dTA6bZzZOanBJ0NSyn1PsGTeVk6cDqaVSzPU8vVmfP37DKszgfcNG2GuKScmXg8jU12ebrtL21vH6ZexxZD8AYsaVGZRQgtqFzFAP18+9AsYY2JujoWlBRamppgaerFmyXuc9k6ibMY4UBGA3zznjXnDuhRN184ljpszm7DEaj97+pmp5mVDCsQ/wABrq8xDVPmL7fRr14NjFbbhvqs7yYPxZCF0Ywsq3ZqEh2tjdEM30LzcTSZ6raUJzzm6aj0Xrx7Grc11TgxQ3XQlLcmbxV1nYL1mK12TO85KvL24jInOJ4ixb4dDl9aUuzmfFWV+ZXnz7O+2FHNzEhW6PKXz5LE49jDj+rZzxNXtRecKqnQq5j7Luvdmc0hJGrRsQdNW7WlTOX1zqog9P1DlzAAu9vJm9eE36Cu2j5li25gp/ppbWGJeqCD5DXTQ1tb99MZwOSC9OUC/en3YG9uGzY930CW7lmUqcq99jO4+mF2GTlw9MorSOcoTFPv7nQPsu/gWixa96GSv7EURjeeR39gf2ZhhDpVSEk25Gyt+6sra4PI0adON/kM6Uj41A83jlMcFcSOnPE5xvLm9exnTJ60gqPl8pg/pwvdVU5t6Sry6tZ/1c8Yz7+gbxX46GqdJQ+jdyFaMkPKFEGXwWxGJ55XTnNgwC8fNzyisSPgb2RfBrIAaUa+f8dBbjbLtBjN+eBtK5KoZrfClS43PlccB5bMvJklQLiO9OsGcTZEMmtKVdJ3upTii40BbTzuTpkGZiHrEWqcNBFbrzODuDRSBcxgeT8OwLmubJqmQ8+TUeWTftaDMJ2c4iTfXz+Nf7Dtqmmf1L0oEvAzEzCrlHg45FX1jOeve9WVUm3/vtqVhVzdwIKET/RpmEtyn9fYoW29UwKGDNTLpFfvWXMB+SM+U+xrEebK192g8xx5gdvXMW/BLkpTcsTm9aHwv72XbzsOcu/GUhFarODHvu+zvSiz34Pc1D/luWEsiLt1Do1ojSuUy2JUUCc6sY7WZOtI++z4Hf9O7k5OZ4NYV13GVcxUIRT3Yye7ARvRu8blhctOIdufaA20q1Cn28beLdj/L+Sh7WlXL+U3c8jJlGRcBiiD8d0QZFIRvW2p8/kUmCYIgfL2UZVwEKILw3xFlUBC+bamxtvI4kD7yFwRBEARBEAThmyeSBEEQBEEQBEEQ0sk0ScjsUmNO5ynl5FJlTt6T0d9ZRhAEQRAEQRCE3BFXEgRBEARBEARBSEckCYIgCIIgCIIgpCOSBEEQBEEQBEEQ0hFJgiAIgiAIgiAI6YgkQRAEQRAEQRCEdJR3TBBDBgmCIAiCIAiCkEw5oujHJCHtHZczPpTSPldKOz8zWc1PKyfvyejvLCMIwpdBWb7FUMeC8N8RZVAQvm2pcbbyOCCaGwmCIAiCIAiCkI5IEgRBEARBEARBSEckCYIgCIIgCIIgpCOSBEEQBEEQBEEQ0hFJgiAIgiAIgiAI6eTpJCHuyUX+9JGrpnIh+hGXrr5BUk2mkPPS248Y1VT25LwLDlM9/7fI8br/iKD0K5ZLctwvnMU9SjX5j0kE+Prn4ndRLOFzm9v+Gb9EBKfXLOdSkGoyt6KfcutusGoiF6QAPDyCM2zr3JF87/MgIJNPiHvI+T/9iVNN5lw0N0+d49U/2s7/BonIyAw7itybK5eeKdZQEARBEAQha3n7SsKH04xoOoEzuY1o1CI4PqYPq56lTTDUUbu7kqHO53iTo+AtlsszOjHqqDLqlQgKCFSE5xlIwTx7krsAO/bEMGqNOMw71XTuqaPzbDntxx8i9ymMhP/R7RxLF73G82LtCMbt9895oK12geHVR3M+3XbRIOmWCwtP/Y1AX0km53DvRsy8+4Kjvy7nhE8OQ3OZjEtTmjLpbKRqxt+QdJL+NSZyJeOGlN6ytVFRRpzK5Q4Yc49d3ZvidPofrFOmovA4tpcLOc4+Yrg2vw8zL4Qjhb4jNHmx95x17Mb8a7lPfQRBEARB+Hbk6SRBu6gVhQ30MZSpZuSUlhVFrE0xNtdMnox6F6oIl2QUadedsk9O8EgVH0W6XeHmy6wCLh2sjRMIitZRPI/Hf7sLiw4eZteG5SxwGsewwQMZ0K8t5e2ncjrHEb8m1raWfHgfj5ZqDlG+eGdWi50lGaYlSmJXyFixhrmVSJSbK7M3PkpTO55ETFQQsUnaik/OGZmRORYVbCihp5qRTA2DQs1pWMcoeSrK9wnugbn4XtolaVC3KPHqVjRrEsO8ob/xMiaa6Gw/Ih/mhvmI10i79nLC/R9z5fgu1jg2w67Frzz6zAUpmbE5pjbGFP64UeKIUf5AMjPMre2xsdNOmZ1DMQ8PcbLwLEY0MlDNybkY76sc2ruTLetXsdR5BpMnTGDixAk4Oo5n7MBONP9+NMuPPP381Q25N+eP3VIkotpYFdAiQtJS5NtTGLrWCykxkSQ1a6ysNVRvFgRBEARB+FTevpKQrwDGtmZYfIxn4ojOSQWoTBe9QgUpmBzbxfDg3Fmeej/h2p/eFKpdiFvOk5noOIIf29Zn+IYbWTa90NItjk0RZaKhhWGiO+dvv0fHqhqt+0zCxXUtrlM60WNsP5oZp7w/M1FXt7HhSuDHWnq1giaUNzP+KyBPuMSUZmM4Ga6azgE1/QKYJAZx48ReNq9wZuJgB7oOX8fdbD8jBs/XxfjZoYIifFSRAngWUpH2zc1UM3JCHRMbKwwVz4KubWXRzMmMHTWBFcdOsXPGMAYP7s+IeWs5cN07i2A2jD2um0ltsRQT7M3DK3/yMvI5W52GMWD2VXi3lZ/q1qOby1Wy+1oynZKULC7hts2R7g596dfYivrOtwnWKIK1ujqVenSiXEq+mLnEWJIMClFQtVEkjyW0b7McN0kTrSKKfcA4N1lqMMdXX6axy2iq6KpmJZOIUCQul/c60brWcA69Vc3OQNfCgvxSArqFK/Jd12FMnb+IhQsX4TJ3AJXk6jTddJKtQ+z/2n6Z0SyKbdByOk64gJpeYczM4rnlbUz71jbI4l8SXKIlzSxy850EITsSYd53uHBsP3sPHOX8XV/CJYkADw/e5aYORBCE/ydywt/44uPj8/Hh6+tPQGAoUX+jlbfwdcqDSYKciDee3Lt8kr27b+DpeZ6V0ycxYcww+nbqQM+Fp3mb7UlHDTVNzZRAXArG/dZZLpx9wNskUyq26MXoGfOZO6Aqlfud5JBTXdJViKehrnhFL19KhqKuaUS5xp3p0LwOFWxN0It5h89tNyKMTPlc/KlrEagIfLd/DIjV9fQoqHh8DNFkmqjlN8AoXUCZRvh9ts0cyxjHKTjNnIvz/AUsO/SIoPAw4g1sKVe2Ab1/3cGelQOoqozaMyXx8ulTgiJu8Sh/W7qUkCEF+vNSmR1F3MO70HfUz5/yTgjl6u8b+TPTVkNxeN65x9vEJD7c384M5z8ILVmfNj8MYfrSRYxo05hWvXvRtVVDGnQYyugOJbMMZoN39qFDz1706tWLYTPWcPCGL+oVymFX3AGXzQc5vm0b227f4/CkusnJyOeoyQwwKGCAXefZbNmxiiH1atL4h250bGLKs0B7Bna0yvIqSdjbt0QlScQ8f8pj1XeOefoIb7OiKONotfyGGOSilERdWcyUY9okHhlP396KhGXAYEXSpHgMGc60lfu5ERDIa3khbLNKLHVtadytJ11a1aNSCWMivV8QJn/DKUWi8LjVr6zsbY++6q1Zk2Hz4yh6ldJEQ10DWdR5HlAaLbeHBHq6o1ezBdYiRxD+JdHPDzGvdycGLrvIK4yxMNHg7YVFdKtRlnr9d+Aer3qjIAh5SALRIT5cWfoj5UuUps3IZWzatZedm5czfWA7GjfqwGDnPdwXWf43TXnv5eT7r79//x51dXWUt2PO+FBK+1wp7fzMZDU/rU/eI3lzaPYc9gZbULZMcUoUDuTUnyWYu6grFqpoMyefGxP0gJUjpvKqtB35NJN4k/ADK7o9Zv7pekzvEceB32+jWbYalRpVp0SGKFYe/IKHTz3xeeHGlc2rOV2wFpUL16KL5XGcHhSniZUO6u8e8UBejtplLTAv3YQeXauR0sgmE5EHGDz+A4t+64kyDo+7NpVBF7qwbmqllOQidD0d+8L6P/orTq+Zk+LiSNSWeHT5Ibb1a6N3byGTLrdl0SgbLo7vwIlm+1na4nNNWyR8XCfxi99r/ILLUsM6gaCr23D//hR77VbgsLWgYl44weExxMYEcvPQW346f5HJlTKG+FEc33KYMm0TcJ6ozYAyrjirrWSU9k5OB2jw/uxRgroMp0txc0wsSlKlhg26kgztjAGp5M/iLtMpuWUzzWM8uHfpNIdP+FJuUEXO/SLHZWtbjrdvhfusayyomnmaERP4HLdnHjx/9pjTrmtxs62AUVxRei6bg57LMAJnbad/6CIG767Hsv6mvDMtgc0niVg0Zzftx+L7SMa0vIJ1/nCKTHal6ekRHOq0C5eq3sz6cQW1drjSQi+OQP9gdK2LJG/HTEVdx3nkMapMcKCIlgm2ymRS9VIqyWcxnWfYsm1LpxwE+4oc8cxipu0PwLrDaMa1tM4y2fko/B6/u6zjRoQeBgnu/OF6B4sRY+hUxhQjI2ssA05xtUwH6sR68fTRHa7fl9N87nL6fPZSy/8fZflW3gpe+DLEPd/GkE7TCR24n20jK6fbp8OuzuPHEcGMvbKMllnVxAh5jiiD35boM+Op/v06Km3wZXuPgqq5EsE3XBnVZwJnjYcrkgdn2hQRNUvfitQ4W3kc+FhHmvGgkNlBIqfzlLKa/1kyW9rP2sS21c5MGdGPrs1KYaavj0G6ODGSl96vMmkiFM7dTVMYOmIc8zZexDe2IGU7TmBEWRPKNDLh0G93KVDBGA2janR1KIPn2snM2/740+Ywsa944haIdqlmNKtWlx7Ov7NjzTDK6pTgx+nL+HXZEkVwXp0PMRUZPX083RsXo4Bq0UzJCmJUME1VtEwDDfW/Cpv8dSDqVkU+23xEpq2tSChkJF7dwK+XIlRzITH0ItfVejL0u+zavssoZBqOn3ovZk0ZwvjpUxnSvB4Vyhrx2C0/3w/5gZ/HOLNitSvrVo+lTYP61C6VyRrFveCBVxyGaomKCS3sBu3ht84GFK7TmibVSmCqr0lS1Gue3flTkUzMoU9dRSI1eBc+GSsiEt4SFObGvnEjcVp+mMcx1jTp05MGJZpSRzrHkRvr2ZXYlz6fJCl/SQhx56GXHIuy+Xj7qDzDNh3m1PEV9CgehFeCJeXyv2L/vF+45r6FH6qWZORuP8VhL6NIfHzfJmfKMvOmLFrRnvin59nvZkAx320snuXC7kM7mdvbAYdu3Rkx5TdOe2VVqxLIyUU7yD+kK/lC9CmTmiDIn3Jw00lSB+mKfxVAgplZ9sG+imG92lgWKM+PzdInCDGBj7ly0+vTcmBYhV7z1rBm2XDsE/LRZlAnSpm+wy/UkCrNi/LcPwJN7+cEPtrHtoD6zN64LM8kCMIXRvJm18wp7MnXkwn90ycISgXrDmNcmwJ8EBWRgpBnqWlrofXJ1XIZJrWGsWLFMKzvLmPc3MM5aMEhfI3yXHMjKSoInyc3OX9kNxvWn+Lq6fU4jhjBiGFDGTKgNw4duzFw7FzWX8k41qYhVfs4s3rFEuZO6k41a3NsbOB9uR408dvL+47zmdDIMjnQkhVqxOR1c6gc5c2HlIU/0rRqRO+hvWlXz46C2orwMSkS9ys38f2QiPzkZKZfiEbd0JCiFqaKIDCaUxMHs/aTKDgtRQD24T1hUaH4PbvHrSd+vLi+hXmTxzB04AAGjN9MlE1Jsmpt9BdtylQw4u3zEOTyBKKD3Dm5ej/emi/ZNmMSjiP78/OwJZzLYuQbDUMjiliVw97WOCUhUdNBV/YQjwLVqJbfBCsT1RokRRGdL20/kL8EH1/MiQg7dEhMvvwUfG4NTrNdWHfwDm8STWi2eDtrxk7CqW9zmvZaysGr7txb141iGSNiuS/BVv3pUd+OyqXzE+XxB5MGr8VT05IOrWMY2/oO7ZYNptRnImmDcu3o168tlvcu4GbswaqfHZiy5wmRkXe5dTmE9/EWNJ93hYsrapGv5BScfiz6aWAecIzjbvqYK76rhkwdjfKDmFDiKZE1GqMZb0qNnwbQuUsvZv++gx279rFn2xx+KJ7ZSsXw5PdfuFF5HEPKxyiShcGMmTCBCY6OjB8/jgHjdvFE1bEiwc+fRMuimW9vKYbwQD883e5x49Jpjh7YxZZtt/G7sw7HUSMZOVLxGDGMwX160GfUfNZu2s05v8y2dwS3Vi3BrcVsmmrGUrHXdH42vsHJY7u4QQd6D+pDQxMtLCpWxTal444g5Frck338fjiYUk2bUjnTKwWGNOjWnpLZX/wVBCEPMm7Yl54NNfHcu40jWQ7yInzN8kaSIL3i7OIBdOrUncFTlrDl5B18gp+yz3U3fjHvUbMfytJVq1mzbjM7Dh7jxB+ujKxnqlo4E5KcuJC7HNpynVc+l/Eu4UBLgyf8sWUF8yd3p1aTaVzRrK0Iutpn3UxICiforQfX9x/Ex7gk6pHqmBfVR13ZvlZDC2PDAoqgM57Y/MUobZZ1NCsFPMHj4n5mTFnAxoMXefAqCl2bOnTqP5GFa9axcuZUHOpYfBrAJovE7eBSZkybwWznBax+YEixsOV0aD+Ng1vXcKnkICYO7Mf4WQtwWb6eLavG0SSrS4JaimQlPpYPAY+4eO0FcUkaxHmEUrRFI+QnJjF+z4uUIV6lSOL0LTDLpHJZQ6sGDt2rKAJcZYqgjkXb6ax1HUthn2NcvXSSfZuW4zxtHCOGdqbV4gtZDtEa5/UUddtqaL94io9JNToM+p76jRtQU18iQl4Ac3M1tFV7ZvTpNfx2N4tBZsPPs+lpXUa2/46xK/qjdfYoD27dQ7MqnD7gRwHb4iSdOota9z4ZOhGnCD60DXc7RRKhCGLUZYrfTfJg391ijJrSjwG929PAzpDsxwCS8D21mcvmQ3DqaINMy5ACNvUYOG8Ri1xcWDi6CXXbtKWxScp73/h/oGCRT689hd5ey5heg5g035XtRy/zwDuMBF0zSlZrQqsq9tQZuoTly5ezfMUqXDdtY9euHWxxnULbohm3dyg3N7hwpvBIZrX0x1e/KTWN1bDpNpmaUbqUko7wh4/iPW5GdOxgm8V+JwjZkQi+c5O7MVrYFs+6kkO3XBUq5KRdnSAIeY+mNRUqFkcj4j4PHudmsHfha5E3kgRZEZqOX8eBA9tZt3whM8Z0wtTDj7q/zKVFw97UcXPhtyc56W4v5831zUwfNZdzsdVp378TTVv+RAONm5y6G4JBmXLkewudncdSL4sWOvKgBxxcPJzuvedzLsiSWj/8TKsyGgTEyrDQTEL+wY8Hj3zwfXiIX5wc2R5tT6XPtbctMZj9j8+xefkiZk0Zy8BmpbCwKEGZ4uYYKCI0/dqdsHywgcuZDqNqQLmOY5k1dxbTp0xiwpRB1IgJo8K48fQfP5CCVy7xLr8Z+jlpLaKWiNexX1j0+0W83oTyLl4No9ptFElFfmr370rCzI7MuBSDFBIEBS0zbf5UsM1IhlRThgOKJCG1yZSsILqaxegwbzEzRjnQsccEFgxvT4tSdll0CFcEypdCsWxZnPjwEALcL7Bt2SZuhEiE3FrL+ve9uXygJecGDGDTowgSEzxwe5rZGEkx3NtyFsshvTGKj0Nm9B0z145C81ocdeY5UdltPaeCvDl4yYheDiUyDYaN2vzK3lHKkZ7UkCmShMSQCOy79Mb+45dPRIrw5frOFTg7DmDI0kt82p87Hp0KXRnYQnWlQpZfkXR8ICoh+UVFQuSPZslSqt8zhufeSdiW+HTwWqPqA/ll2xZcl81n5sRRDO79Ix1aNaZ2ZTtsFBlbTLSyiVd24vA8e4LAKmOY2tkOPK5zNtwAszfHmDZkJfFt+uE4tiYvf/+V0/LaNP9Mni0In5dASEio4v8a5NPPp5onCMLXRYZB/vxoJIYSGZWTc5DwtclzzY2UTSVuu87ncuWpTKypCEjVjWg3og43prlw46/m+J+SXnJy+k8M3aNOh1kL6VZWK7mtuXInN6v5E0P6N0P7wj5Cuy1hrGos/0+F8ujsMR7odGLBxoU4VMyf8hkxz/FSs8E0OgzP03u59DwMDcuqfN9vNLN+rp1pQJ1M/pTVAwaxzT1NgpOYpAg909JAx+t3lh8LVE1nJRq3zbPZZzIap3qGJOnXZexgQ/bM3cbzHAwLq66nT+mWI5gxeST9OlfDQFJDO7W63qgZY8a3oYBiMv7VGzStbP66j0OmlN9A7a+dJ/QS66c4MnvNYa489ua9IodQBt2Z1sJL3hx/a8uPlbXRMylKqWptGThtAxscS/MqvCJjRzfCtJwiQdjYgRdTG9N47Br2Xr2vCK/Tkz/dwwndn+hfMZYPcWpoKjdUzD3ORNnzfTFbunXVZfuoydwq2YdWWWxumbU95ZVXgZKSUNdQrK9ZdWqWCuHsiik4OjoydvwcdrtFE5dkTpX2wxjTtUImnZZ1MbdIM6StYm/QCLvLQcU+7DRuCAOmbeSi94uUm+dJoXgG6GNXPDd9ANTRlKmTlKPjsza2ZQvjt20CA0eOZ9DYk2iZPmf34pU8tGtOLauCaNp0pknkcaIatcMys8xJEHJEAwN9fWSJCURHZWy0KQjC10EiPDyMBA0zjIyyv64ufH3yWJLwjisrZnDIZDhzupX8OLSoeoleTKx3lj7Dd+CdVbM4mRUtZ+/j4C8/UyWTYYKCz/zGCbMRTG5h/jGgk3te5KJH2isURlR1mMqs4d9hpamBnq5ucpIghdzjvYEeLxJ+Yo3rdEZ2rYqtZTFK2lSgzPMF9N/krShKmdC0xjrpBU/f/xXhSYqTavr3ylDTsqdq1SwbPilE8WTrbDbFdmfekMroSFLyZ2iX68PYWjeYOvMk2d2PTV1TC82PncmVBT8enXypm1+Gbd8Fit9Yl9iXkYrVNk0T9GYmSfFfSgqWTKcIpcuXxEL7PX4vw0lQ7FYa6W5u9pfwS8eIqeeAvqc3Ng0bUdq6BJaGhdC5u4gJu/2T+4hIvqdYszuCLluvcGD9esbX0Un/m8mfc+RKfrr2roJeYiRR8VrJCUnouVPI63XCRvFP65WpR6HbtzGqXeqzQ9Qm+xBLwscrI1Y0HeGMi4sLSxeNpE3tOnTq0YWW9SphV8Qw64QwVYwnPn7aWFb4ji6DprLi/HtCNnfAPPm1R7ip2VM5q7YZmUrkQ4wcjRzmFTLLRoxYsp61U2tiUXcSq6e0QiuuBhOG1Uy5shP1gCsBGopczSOTzv+CkFMyLO3tKaETy7NnbmJfEoSvUZwn9+76kGhah1qfbTIhfK3yTpIQ+YTd85Zyr8w4ZnW1yxCMaWI/dD7dnvSg3bAt3A3KJiJWSDu4kvTqEKsuF2NQjzLpPjcx9A433LKuBUtMVAb3CQSd9aJAk1K4nTnFywytntSTPpCobphFUK1JIbPiGCmr6FMlJCKlWTdlwB4br4+JSRZRoBTAxVXOHJA5MGNwjZR7BigSjcTkz5Bh03ka/RIW0cfpKF6fO1Mrlnl97yCui6YyavB4liuC13z5M651HM88ErCyyyYiTZQUoWvqskmK76+PZbkGdB42jxXTv8dMkT+oq2dS6xB1m6O+lfi5UQCHT7mjXzKWS8v34y/JKNHTic6Jj3gVLxGoYU+v+i8YVqM7R/U7M+7n2ulHTtG0o9PAjpRUrmZCNDGJWmgnebDvZiE6tlFkiJIf++dswWz5XOJXL+NmVMpiWZGiooiT/Uu1JFqFqdS4PtXq1KSCXRGM9OKIVm0Xufcd3haugW12AX+0OxdOP1b16ZAIDY1DVz93RVX+XoOCiecY274jG+LLUChOuRIRXFu9A+2Re/gxZBNbPLMvR4KQFd2aP9Ctji7Pj+zjdBZDn8Q8v8XdXN1RXhCEvEEi4OQ6tl7XpFb//rTJzf1Wha9GHkgSJN5e28biFZcwdJjJyKZFPoafUmQMsamV8Ho1Gee6BPtnG5g8aBBTlmzk3IsMEbvkz9ntB3gUqqxpV0MjubI7jtsbT5CvS8dPbiAlhb/mrnvGuwJLvHrmTrAk5324Mrr055CnBR3qWGBioYdO2ng/4hWPXqhjWzrrqmFdnUSiP6RpK6IMsKUYAr3duHv1HMcP7uGwh4RpJkm6POAKWxZv51WN8Ux3qEBqNwopMpJI1XNk5rScvpA2/osZ3Gckc1du5ZzHp1GxPCCAkIJladbDkcWui5k9qSNlUjOm0Icc2LCDay+9uOmtTRmrz0exUtA73queI70hOMqU4hXLYWueL7lGP1EeT6KyI3BG+tXp0ac2YVtmc0OnJoZmHRhU4yrDB7lyO6I8jpucaaAbh/vJE3hXcWLbdE3OnHyp2CKfkRRKWLQmsSf2El7/ZyqrB3Dml5X4tVnAxNY/MrLRFXp2W8yfr7Lu0xIfFkG0hmZyYYiLCsL/+UNu/nmKP/af48HNIywdP5SBfRzo2KImZes7cvxzLcNkReg2czx1U3cJKYTzK2ezwy0Cr+M3yf9drSz6aqShV5rqRbzYtGCtYh/34bk3GBtnl1mkp1m8JtWt9SnZZzsbOwbiOmcD5/Yv5YjpMMbUs6JFZ0N2TFjHk487kiDkknYVBs6bRtMPW5g0ehVXA9OXsfCHe3C9FIPVZwZ2EAThv5UUG/dXnPVRDL6nFjJozA7Uui1l1bi6H+MP4duiDKOT66TDw8OT25Erb6KQeiOF1JurKWX1Vynt87Symp8q+XW5J1euQ6UGJT8ZZztoV3eaX/6ZK6tafHwtu8+Ue//BDMcVnIxpxNqDTlTTlgg8uxzng95IikXV1ROJ+xCtCNzDCX4djE4bZ3ZOapBudA657wmWzNvKidPBtHJphlq+3oyvf59hdSbwvmEjzDXlxMTrYWSqy9Ntd2l76zj9MmstJH/AogaVWZTQgtpFDNDPlw/9AsaYmJtjYWmBhakppoZerFnyHqe9kyibNg5UBN83z3lj3rAuRdNdVonj5swmLLHaz55+uUjtpUD8Awywtso8RJW/2E6/dj04VmEb7orfPXkwniyEbmxBpVuT8HBtjG7oBpqXu8lEr7U04TlHV63n4tXDuLW5zokBqTdmSUPyZnHXGViv2UrX5I6ziiTx4jImOp8gxr4dDl1aU+7mfFaU+ZXlzbMfFiXm5iQqdHlK58ljcexhxvVt54ir24vOFVSHtJj7LOvem80hJWnQsgVNW7WnTeX0zaki9vxAlTMDuNjLm9WH36Cv2D5mim1jpvhrbmGJeaGC5DfQQVtb99Mbw+WA9OYA/er1YW9sGzY/3kGXz7UsS0PutY/R3Qezy9CJq0dGUTpHeYJif79zgH0X32LRohed7JW9KKLxPPIb+yMbM8yhUsrBXu7Gip+6sja4PE3adKP/kI6UzwNnAWX5/lv3WBH+IxJBNzfjPNOFHU8MqNu8DnamWkSHRaJVtjOjhzXDSuQIXxRRBr8VkXheOc2J9bOY+LsHlg270qJaEQwV8c37N968iilMrY696NulGiLP/7akxtnK44Dy2X+bJHyG3G0NU09WZc64Gh+bCWW3jJL06gRzNkUyaEpXLNLu3FIc0XGgraedRfOgNKIesdZpA4HVOjO4ewNF0ByGx9MwrMvapkko5Dw5dR7Zdy0ok2kAJ/Hm+nn8i31HTfOs/kWJgJeBmFml3MMhJ6JvLGfdu76MavPvji0YdnUDBxI60a9hJsF9Wm+PsvVGBRw6WCOTXrFvzQXsh/RMua9BnCdbe4/Gc+wBZldP32gslSRJyftaetH4Xt7Ltp2HOXfjKQmtVnFi3nefJI6fkHvw+5qHfDesJRGX7qFRrRGlchnsSooEZ9ax2kwdaZ99n4O/6d3JyUxw64rruMrZ95FII+rBTnYHNqJ3i6yGyc0g2p1rD7SpUKfYx98u2v0s56PsaVUt5zdx+68oy7cIUL5Ect55PeaZXwgxmoUoYV+RYoZ5fW8TMiPKoCB821Lj7DyfJGTm7ywjCMKXQVm+RYAiCP8dUQYF4duWGmcrjwN5oE+CIAiCIAiCIAh5iUgSBEEQBEEQBEFIRyQJgiAIgiAIgiCkI5IEQRAEQRAEQRDSEUmCIAiCIAiCIAjpiCRBEARBEARBEIR0RJIgCIIgCIIgCEI66ZKEjGMjZzZWcm7GT87uvbn5rFR/ZxlBEARBEARBEHJOeccEEXULgiAIgiAIgpBMWSkvmhsJgiAIgiAIgpCOWmhoaJLyFszKh7q6evIj9ZbMqfNSn6dKfX+qtM8z+txrStm9npm/s4wgCIIgCIIgCDkjriQIgiAIgiAIgpAG/B+n/K3xlWaISgAAAABJRU5ErkJggg==[/img][br][br] 如果算法的查准率高而查全率低,虽然分类效果的可靠性高,但对新的语句进行分类时很多正确的类别不能识别。而如果算法的查全率高而查准率低,虽然对新语句的正确识别效果很好,但分类结果中错误的数量可能比较多。由此分析,单独使用查准率和查全率中的一个指标来评价分类算法是不全面的,需要综合考虑。