Eventregistry

**Updated**
- Text analytics: categorization API now supports additional parameter for determining the taxonomy of interest

**Updated**
- Important bug fixes related to article and event iterators. The existing classes had issues when accessing the data from the last month as well as historical data.

**Updated**
- Added some utility methods to the `TopicPage` class.

**Added**
- added `TopicPage` class. It can be used to create a topic by specifying keywords, concepts, sources, ... as well as their weights. You can specify a threshold and receive only articles and events that match enough specified conditions to reach the required weight. Alternatively you can simply sort the results by relevance and get the top ranked results that are most related to your topic page. See `TopicPageExamples.py` file for examples on how to use the topic pages.
- added named entity extraction endpoint to `Analytics`. Call `Analytics.ner(text)` to extract named entities.

**Updated**
- sentiment analysis now supports two models - vocabulary based as well as a model using neural networks. Choose the model by specifying the `method` parameter. Possible values for it are `vocabulary` (vocabulary based sentiment analysis, default) or `rnn` (neural networks based model).

**Added**
- added `blog` data type. Various methods in `EventRegistry` class accept it, such as `suggestNewsSources()`, `suggestSourcesAtPlace()` and `getNewsSourceUri()`.

**Updated**
- `QueryArticlesIter.initWithComplexQuery()` now accepts also the `dataType` parameter (by default `news`).

**Removed**
- Removed the parameter `articleBatchSize` from `QueryArticlesIter.execQuery` since it was not useful. We are always returning the maximum number of results that can be obtained with a single query.

**Added**
- Text analytics: added `Analytics.semanticSimilarity` API call. It can be used to determine how semantically related two documents are. The documents can be in the same or different languages.
- Text analytics: added `Analytics.extractArticleInfo` API call. It provides functionality to extract article title, body, date, author and other information from the given URL.
- `dataType` parameter when searching for articles. Event Registry is now separating collected content by data type. The possible values for data type are "news", "pr" (for PR content) and "blogs" (we will start indexing and providing blog content shortly). The dataType parameter can be set in the QueryArticles and QueryArticlesIter classes as well as in the `EventRegistry.getNewsSourceUri` and `EventRegistry.suggestNewsSources`.

**Changed**
- changed params in the GetCounts and GetCountsEx classes.
- `EventRegistry.suggestNewsSources()` and `EventRegistry.getNewsSourceUri()` now also accepts `dataType` parameter, which is by default ["news", "pr"]. It determines what kind of data sources to include in the generated suggestions.
- `QueryArticles` and `QueryArticlesIter` classes now supports additional parameter `dataType` that determines what type of data should be returned. By default, the value is `news`. For now it can also be `pr` or an array with both values.

**Removed**
- Removed the RequestEventArticleUris, RequestArticlesUriList, RequestEventsUriList due to backend changes. Use the equivalent \*UriWgt\* version of the classes.
- Removed RequestArticlesUrlList class since it is not supported anymore.
- Removed `QueryArticles.addRequestedResult()`, `QueryEvents.addRequestedResult()`, `QueryArticle.addRequestedResult()`, `QueryEvent.addRequestedResult()`, and `Query.clearRequestedResults()`. As before, a single result type can be requested per call so the methods are not usable. Use `setRequestedResult()` methods.
- Data model change: We removed the `id` property from different returned data objects. Although the documentation clearly stated that the property is for internal use only, users commonly used the property, which caused potential issues.