Feed aggregator
Maximum number of concurrent sessions in multi instance database
Index on XMLTYPE with XPATH Expression including a XPATH Function
Cannot Upload git-upload-pack error while cloning Azure Git Repository
音楽専門学校におけるインターン制度で就職
音楽専門学校におけるインターン制度で就職
音楽専門学校に通学する方々の動機というのは人それぞれ大きく異なります。プロになりたいと考えている方が居る一方で、業界に入り何らかの立場で働きたいと考えている方も居ます。数ある進路の中で音楽専門学校という選択肢を選ぶのは、独学には存在しないメリットがあるためです。音楽に携わってお仕事をしていくという場合は、どのような形態であっても様々なスキルや知識を持ち合わせている事は必須です。
専門性が高い上に、感性も大きく関係するためインターネットや書物を用いても知識を得る事は困難で、指導員から直接的に学ぶ事に越したことはありません。歌手や楽器を演奏するといった事は、特に感性により左右されるので座学でプロの腕前になる事は不可能です。そして、指導員の多くは普段スタジオなどで録音作業に携わっているプロのミュージシャンです。いわゆる二足のわらじを履いている状態なので、指導員を介する事で業界に足を踏み入れるきっかけができるという可能性は十分にあるのです。
しかも、音楽専門学校としても就職率を高めたり優秀な人材を業界に投入させるという意味合いから、業界と学生の接点を数多く用意しているのです。また、音楽専門学校は就職希望者を中心に一定期間提携している企業で働くという制度を導入している事もあります。刻々と変化する音楽の業界で活動することにより、学校では学べない事を学べます。卒業後は、インターン制度を利用した企業にそのまま入社するというケースも珍しくありません。
FORCE_LOGGING in Autonomous Database
Find Circular References in UDTs
Select XMLQuery XML parsing error with ampersands
Patch Oracle GoldenGate Microservices using RESTful APIs
In 2017, Oracle introduced the world to Oracle GoldenGate Microservices through the release of Oracle GoldenGate 12c (12.3.0.0.1). Upon the […]
The post Patch Oracle GoldenGate Microservices using RESTful APIs appeared first on DBASolved.
How can we execute a SQL script file in SQL trigger and output of this SQL script execution into the log file?
PLSQL nested procedure hides resolution of an outer procedure
External table in a PL/SQL procedure
Configure of Oracle Data Miner repository in SQL Developer Desktop to work with Autonomous Database
個性的な人材が音楽専門学校に不向きな理由
個性的な人材が音楽専門学校に不向きな理由
音楽が好きで自分で演奏する機会も多いという方は、必ず音楽専門学校に進学すべきか迷うタイミングがあります。答えを導き出す事が難しい悩みではあるものの、参考にする事ができる一つの情報があります。それは、プロの中でも特にレベルの高い人材を目指すのであれば、音楽専門学校に進学すべきではないということです。極めてレベルが高い人材を目指すつもりではなく、音楽に携わる事ができるお仕事に従事する事ができれば良いとい考えの場合は進学すべきでしょう。
このような判断基準が設けられているのは、音楽というものは感性が大きく関係している分野であるためです。感性が大きく左右する世界だからこそ、大勢の人々と共に同一の事柄を学んでいると技術力と知識が平均的なもので収まってしまうのです。人とは異なる個性を有していたとしても、個性は消失してしまう可能性が高いです。つまり、大勢の人々とは明らかに異なる魅力的な個性を持っている方は音楽専門学校に通うよりも、荷物片手に世界を歩き回る方がより一層魅力を増幅させることができます。
音楽業界を見渡してみると理解することができるように、第一線で活躍することができている方というのは、強い個性を有している方々ばかりで、音楽専門学校に通っていたという経歴を持つ方は多くありません。ただし、現在人よりも秀でている点が無いものの将来音楽に関与するお仕事を行いたいという考えを持っているのなら音楽専門学校に行くべきでしょう。
table with 900 million records with 2 clob fields and weighing 5tera and without indexes
Oracle Edition Based Redefinition EBR Table Data
The challenges of securing data in an Oracle database
Posted by Pete On 11/05/22 At 10:04 AM
Analyzing Brand Reputation with Oracle Analytics
When Italy used to qualify for the FIFA World Cup, I was a big fan of Oracle Endeca Information Discovery and implemented several innovative projects with it. Endeca allowed users to upload and combine diverse data for agile discovery on structured and unstructured information, transform and enrich data with a powerful visual data integration environment, mine text in multiple languages, and identify sentiment through natural language processing within the same framework. Unfortunately, it was too ahead of its time and didn't achieve the success it deserved.
Two World Cup qualifying failures and one Euro Cup triumph later, the time is ripe and several features seen in Endeca have been included in Oracle Analytics. In this post, I'm going to illustrate how to use Oracle Analytics to easily perform brand reputation analysis without machine learning or marketing experience.
Let's begin!
Identifying the Brand to AnalyzeAccording to Marketing91:
Brand Reputation is the perception of the brand and how the particular brand of the company is viewed by customers, stakeholders, and the market as a whole. It is the culmination of ideas and emotions that a customer associated with the brand with the customer service experienced during the purchase of goods and services, whilst using them, and after-sales services provided by the company.Since Italy can have a chance to win the 2022 FIFA World Cup only in videogames, I'm going to pick eFootball 2022 - a free-to-play football game - to conduct my brand reputation analysis.
At launch, the game was panned by critics and players, who criticized the atrocious graphics, lack of content, laggy engine and finicky controls. With 92% negative reviews, it became the worst-rated game on Steam a day after launch, and the lowest-rate game of 2021 on the review aggregator Metacritic.
Konami - the publisher - apologised for the game's many issues and said they would work on improving it. A major update was released on the 14th of April 2022 on PC and consoles: would it be sufficient to raise the reputation of the game?
Gathering the DataPeople love to share their thoughts and opinions online. Due to this fact and the rise of social media, consumers have the greatest say in how a brand is perceived nowadays. Online discussions, reviews, posts, and articles about a brand can make it or break it.
I'm going to focus on a single social network - Twitter - to start my analysis and understand what people think about eFootball 2022. Unfortunately, Oracle Analytics does not (yet) allow users to gather data directly from social networks, so I decide to use the Twitter API v2 and search-tweets - a Python library for the v2 search endpoints - because I'm familiar with them. I don't cover the details here, but you can trust me when I say that the most difficult thing is to get the developer account application approved by Twitter!
In order to have a set of data suitable for analysis, I'm going to consider only tweets in English, and ignore retweets and quotes. I'm also going to exclude tweets with links because I noticed that most of them were actually spam. I can then retrieve all tweets posted over the last week related to eFootball 2022, and its official Twitter account @play_eFootball, with the following query:
(to:play_eFootball OR @play_eFootball OR #eFootball OR #eFootball2022 OR eFootball OR eFootball2022) -from:play_eFootball -is:retweet -is:quote -has:links lang:en
The response is in JSON format, which is not directly supported by Oracle Analytics, and contains a mix of tweets and users information, so I'm going to convert it in a couple of Excel files - Tweets.xlsx and Users.xlsx - using a simple custom Python script.
Tweets.xlsx contains the following fields:
id
- The unique identifier of the requested tweettext
- The Actual UTF-8 text of the tweetauthor_id
- The unique identifier of the user who posted the tweetcreated_at
- Creation time of the tweet- Public engagement metrics for the tweet at the time of the request, such as
retweet_count
,reply_count
,like_count
andquote_count
Users.xlsx contains the following fields:
id
- The unique identifier of the username
- The name of the user, as they've defined it on their profile (subject to change)username
- The Twitter screen name, handle, or alias that the user identifies themselves withlocation
- The location specified in the user's profile (freeform value)- Public metrics about activity for the user, such as
followers_count
,following_count
,tweet_count
andlisted_count
With the gathered data in Excel format, I'm finally able to upload it to Oracle Analytics and use the data preparation features to enrich the datasets before visualizing the data.
Focusing solely on the numbers is a big mistake in brand reputation analysis. The high number of mentions does not automatically mean the brand is doing great, in particular when the vast majority of those mentions are terrible reviews. Understanding the general attitude towards the brand it's a crucial activity, but it could be really tedious if you have to do it manually.
Luckily, the Analyze Sentiment step comes in handy: in Oracle Analytics, it can be added to a data flow to automatically detect the sentiment (Positive, Neutral or Negative) for a given text column. Figure 1 shows how easy is to configure this step. All I have to do is to select the column to analyze (text
) and provide a name for the new column that will contain the calculated sentiment (the default is emotion
).

The next step requires to save the data in an Oracle Database or Oracle Autonomous Data Warehouse, for instance by adding a Save Data step to the data flow and setting the Save data to property to an existing database connection (Figure 2).

Having the data in the database allows to display the Database Analytics step in the data flow editor and perform advanced analysis and data mining analysis. In particular, I'm going to use it to break tweets text down into distinct words, count their occurrences and save the results in a new database table. Once again it's very easy to configure this step. All I have to do is to select the Text Tokenization analytics operation (Figure 3), specify the column to be tokenized (text
), the reference column(s) to be included in the output dataset (id
), the number of top tokens to return and wheter to use a special language lexer (Figure 4).


Tokens and occurrences will be fundamental when I analyze the data to display Tag Cloud visualizations. It is a great visualization of the things people usually associate with the brand, and it can be an indicator of the brand reputation in general.
Since all tweets, users and tokens data is in the database, it's a good idea to create a dataset with multiple tables to analyze the data (Figure 5). This solution allows me to achieve better performance because all the joins will be executed at the database level and only the tables needed to satisfy a visualization are used in the query.

Now that data is set up, it's time to conduct a brand reputation analysis to identify what works and what does not, and optimize the brand strategy accordingly.
The Line visualization is the ideal way to display brand mentions (i.e., the number of tweets) over time (Figure 6). The number of mentions fluctuate with the higher peaks on the 21st of April, when eFootball 2022 Season 1 started, and on the 25th of April, when Konami made important announcements related to the game. The decreasing trend of mentions should not be cause for concern: most of the tweets are created in the afternoon/evenining while I gathered the data on the 28th of April during my lunch break (Figure 7).


As stated previously, the sentiment analysis is crucial for a successfull brand reputation analysis. I like to analyze it either at aggregated level with a Donut visualization (Figure 8) and over time with a Line visualization (Figure 9). According to the Analyze Sentiment step 47.66% are positive and only 21.64% are negative, a big step forward compared to when the game was originally published in 2021!


Users have a major impact on brand reputation and a good brand reputation analysis should take into account what the most relevant influencers are posting. I usually display a Table visualization of the top 10 users by mentions, by followers count and by engagement (i.e., the sum of mentions' retweets, replies, likes and quotes), and create a data action to a detailed canvas for inspecting the user's content (Figure 10).

Finally, there is the Tag Cloud visualization which clearly shows what people are looking for when mentioning the brand (Figure 11).

With the fundamental visualizations clear in mind, the brand reputation workbook can take shape. In the Overview canvas (Figure 12), I like to present aggregated data such as the total number of mentions, the sentiment ratio, the sentiment over time, the number of mentions over time, and the top 50 words by occurence. I can restrict the analysis to a specific emotion, a date range, a peak or a word, and see how the selection affects the other visualizations.

The Influencers canvas (Figure 13) lists the most relevant influencers alongside with the sentiment ratio and the most used words by mention, followers count and engagement. Again, I can restrict my analysis to a specific user, an emotion or a word, and see how the selection affects other visualizations.

The last canvas, Tweets (Figure 14), allows to fully inspect the content of the gathered tweets and related public metrics. Since the data in this canvas is at a very granular level, I prefer to reach it through data actions from the Overview or Influencers canvases. This allows to preserve any selection previously made (i.e., the search context) and display the details only for the tweets that I'm really interested in.

Once the data is available in a format supported by Oracle Analytics, it takes less than an hour to build a workbook like the one described in previous sections. It's worth saying that while it represents a solid starting point for a brand reputation analysis, there is also large room for improvement:
- Manual data gathering is not the way to go in a production scenario, and the process must be automated.
- I gathered data from Twitter, but there is a wide variety of platforms people use to mention brands and all relevant sources should be monitored.
- Competitors' data should be collected to evaluate the strenghts and weaknesses of brands that are deemed to be a potential threat to your business.
Brand reputation nowadays is a must have intangible asset for any organization since even the small error in service can be visible to the entire world by a single tweet by that unhappy customer, and impact the business.
If you are looking into brand reputation analysis with Oracle Analytics and want to find out more, please do get in touch or DM us on Twitter @rittmanmead. Rittman Mead can help you with a product demo, training and assist within the development process.
Pages
